AI Pricing History

Zapier's free plan now includes 100 tasks per month, with access to Zap workflows, Tables, and Forms. Users can start automating with AI at no cost.

Official source; free tier details explicitly stated.

Mistral Small 3.2 is scheduled for API deprecation on 4/30/2026 and retirement on 7/31/2026. The recommended alternative is Mistral Small 4.

Clear deprecation and retirement timeline provided.

Mistral Small 4 is listed as a new hybrid model unifying instruct, reasoning, and coding in a single efficient model. It is available under the Apache 2.0 license, version 26.03.

No pricing or plan details are mentioned; this is a model launch.

Mistral Large 3 is listed as a state-of-the-art, open-weight, general-purpose multimodal model. It appears as a newly available model in the official documentation.

No pricing or plan details are mentioned in the provided text.

Forecasting predictions are $0.20 per 1,000 counts up to 1M, $0.10 per 1,000 counts for 1M–50M, and $0.02 per 1,000 counts above 50M per month. Training is $21.252 per hour.

Clear tiered pricing for forecasting predictions.

The documentation lists available Gemini models and their endpoints, but does not mention any pricing, plan, or feature changes for Gemini 1.5 Flash.

No explicit pricing or plan information for Gemini 1.5 Flash is present in the extracted text.

Cline is a free autonomous coding agent extension for Visual Studio Code, supporting multiple AI model providers and offering features like file editing, terminal command execution, and browser automation. The extension is listed as free on the Visual Studio Marketplace.

No pricing or plan changes detected; extension is currently free as per official marketplace listing.

Replit Enterprise offers all Pro features plus custom seat limits, SSO/SAML, advanced privacy controls, dedicated support, and other enterprise-grade options. Pricing is custom and requires contacting sales.

Enterprise plan details are clearly listed on the official Replit pricing page.

Replit Starter is a free plan that includes daily Agent credits, a built-in database, and the ability to publish up to 1 project with private or password-protected deployments.

Free tier details are clearly listed on the official Replit pricing page.

Cursor Router, an intelligent model router, is now available. It analyzes requests and routes them to the appropriate model based on optimization modes: Intelligence, Balance, or Cost. Admins can control router settings per team or group. Available by default for Teams plans and can be enabled by Enterprise admins.

Official changelog; describes new model routing feature and admin controls.

The documentation references models deepseek-v4-flash and deepseek-v4-pro, but does not mention DeepSeek-V3 as available or updated. No pricing or plan changes are specified for DeepSeek-V3.

The extracted text does not mention DeepSeek-V3 model availability, pricing, or plan changes.

The o3 model was released on April 16, 2025, with a context window of 200K tokens and a knowledge cutoff of June 2024.

Official release date and model specs are clearly stated.

Claude Code is included in the Pro, Max 5x, and Max 20x plans. Pro is $17/month with annual billing ($200 upfront) or $20/month billed monthly. Max 5x is $100/month, and Max 20x is $200/month. Usage limits apply.

Prices do not include tax. Plans and prices are subject to change at Anthropic's discretion.

Cursor for iPad is now available on all paid plans, offering a rebuilt layout for larger screens, sidebar chats, split screen, and enhanced review features. The update also adds an inbox and multi-PR session support on iOS devices.

Official source; feature expansion for paid plans.

New users can get started with $1 in free credits for serverless inference.

This is a clear mention of a free tier incentive.

Together AI and Y Combinator have partnered to launch the first dedicated GPU cluster for the YC community, eliminating the need for two-year compute contracts and providing YC startups with faster access to GPUs.

No specific pricing or plan details provided.

Agent Platform pricing remains the same as legacy AI Platform and AutoML products, except that lower-cost, lower-performance machine types and scale-to-zero are no longer supported. New features include optimized TensorFlow runtime, co-hosting models, and no minimum usage duration, with charges in 30-second increments.

Official Google Cloud source; specific exceptions and new features noted.

The Starter plan is free and includes daily Agent credits, a built-in database, ability to create slides, videos, and animations, publish up to 1 project, and private or password-protected deployments.

Free tier features are explicitly listed on the official pricing page.

Enterprise plan offers all Pro features plus custom seat limits, SSO/SAML, advanced privacy controls, design system support, data warehouse connections, custom groups, dedicated support, single-tenant environments, region selection, static outbound IPs, and VPC peering. Pricing is custom.

Enterprise features and custom pricing are clearly listed on the official pricing page.

Replit Pro includes everything in Core plus $100 monthly credits, up to 15 collaborators, up to 50 viewers, parallel work with up to 10 agents, access to the most powerful models, database rollbacks for 28 days, and premium support.

Plan features are explicitly listed on the official pricing page.

Replit Core includes everything in the Starter plan plus $25 monthly credits, up to 5 collaborators, parallel work with up to 2 agents, unlimited workspaces, and AI integrations.

Plan features are explicitly listed on the official pricing page.

Replit Pro is priced at $100 per month, or $95 per month when billed annually (a 5% discount). The plan includes $100 monthly credits, up to 15 collaborators, and additional advanced features.

Pricing and features are clearly listed on the official Replit pricing page.

Groq offers a free API key for new users to get started, with the option to upgrade as needs grow.

No specific free tier usage limits are mentioned in the extracted text.

Trial API keys are free but rate limited and not for production use. Production API keys are charged on a pay-as-you-go basis.

Trial keys are automatically created for new accounts.

Compass, the intelligent search and discovery system, is available only via custom enterprise pricing. No fixed pricing or self-serve plans are listed.

Users must contact sales for pricing.

North, the all-in-one AI platform, is available only via custom enterprise pricing. No fixed pricing or self-serve plans are listed.

Users must contact sales for pricing.

Mistral AI announced the launch of Mistral Small 4 on March 16, 2026. No pricing or plan details are mentioned in the provided text.

No pricing or plan changes are specified in the extracted text.

Claude Mythos 5 shares the same pricing as Claude Fable 5: $10 per input million tokens and $50 per output million tokens.

Anthropic announced Claude for Teachers, a new product offering tailored for educators.

No pricing or plan changes mentioned; announcement is for a new product.

The Enterprise tier for Gemini API includes all Paid features plus dedicated support, advanced security, provisioned throughput, volume-based discounts, and ML ops/model garden options.

Enterprise features are listed on the official pricing page.

The Free tier for Gemini API now includes limited access to certain models, free input and output tokens, and Google AI Studio access. Content from the Free tier may be used to improve Google products.

Official source confirms Free tier features and limitations.

Google has introduced three new AI subscription plans: Google AI Plus ($20/mo, 400 GB storage), Google AI Pro ($35/mo, 5 TB storage), and Google AI Ultra (starting at $99/mo, 20 TB+ storage). Each plan offers increasing usage limits, access to Gemini models, and bundled benefits such as YouTube Premium and Google Cloud credits.

Pricing and features are clearly listed on the official Google One AI Premium page.

Amazon Nova Pro is a newly available multimodal foundation model accessible via Amazon Bedrock. It offers a combination of accuracy, speed, and cost for a wide range of tasks, supporting text, image, and video inputs, and is available in multiple regions including US East (N. Virginia), Asia Pacific (Sydney), Europe (London), and AWS GovCloud (US-West).

No specific pricing or plan changes are mentioned in the provided text.

Pipedream has entered into a definitive agreement to be acquired by Workday. The announcement states that Pipedream's AI agent-building products and integrations will be combined with Workday's offerings.

No pricing or plan changes are mentioned in the provided text.

Pipedream offers a free account sign-up with no credit card required and describes its free tier as 'generous.' No specific limits or pricing numbers are mentioned.

No explicit numbers or detailed plan changes are provided in the text.

Make has launched support for Anthropic Claude models, including Claude Opus 5, in its AI Provider and AI Toolkit features.

No pricing or plan changes mentioned; only model availability updates.

Make AI is now available for all users as of October 18, 2024. No pricing or plan details are specified in the provided text.

Source is official and the launch date is specified, but no pricing information is given.

UiPath offers a Basic plan for Automation Cloud Test Cloud starting at $25 per month, targeting individuals and small teams. Standard and Enterprise plans are available with additional features, but pricing for these tiers requires contacting sales.

Pricing for Standard and Enterprise plans is not listed; only Basic plan has a specific price.

n8n introduced a Start-up Plan offering 50% off the Business plan for companies with under 20 employees. Community Edition remains available as a standard, self-hosted version.

Start-up Plan eligibility is for companies with fewer than 20 employees.

n8n offers a free trial for Starter and Pro plans with access to all Pro features, 2,500 executions, and 800 AI Assistant credits. No credit card is required for Starter and Pro trials.

Business plan free trial requires a credit card and offers 14 days of full access.

n8n updated its pricing: Starter plan is 20€/mo (billed annually) for 2,500 executions and 2,300 AI credits/month; Pro is 50€/mo (billed annually) for 10,000 executions and up to 13,700 AI credits/month; Business is 667€/mo (billed annually) for 40,000 executions; Enterprise is custom pricing. All plans include unlimited users and workflows.

Pricing is based on workflow executions, not per step or user. AI credits included per plan.

Zapier has unified its pricing so that AI steps, code, and SDK now all follow the same task-based pricing model. The Free plan includes 100 tasks per month at $0/month, and paid plans start from $1.

Official Zapier pricing page; clear mention of new unified task-based pricing and free tier details.

The Roo Code Extension was shut down on May 15th, 2026. The product is no longer available, and users are directed to alternatives. Paying users with billing questions are advised to contact billing@roocode.com.

Official repository notice; no mention of refunds or specific plan details.

Cline is listed as 'Free' on the Visual Studio Marketplace, indicating that the extension does not require payment for installation or use.

No mention of paid plans or pricing changes; extension is currently free.

The continuedev/continue repository is no longer actively maintained and is now read-only for all users. A final 2.0.0 release was made, which included removing anonymous telemetry, pulling out authentication, and bug fixes.

No pricing or plan changes are mentioned; this is a project status update.

Continue has been acquired by Cursor. The open-source codebase remains freely available, and the team expresses gratitude to the community.

No explicit pricing or plan changes mentioned; acquisition and open-source status confirmed.

Enterprise plan offers volume-based credit pricing for large organizations, with the same app and AI feature costs as Pro and Business plans.

Official source; enterprise pricing model described.

The free plan includes a daily grant of 5 build credits (up to 30 per month), 20 monthly Cloud credits, and 4 AI feature credits for trying built-in AI features before subscribing.

Official source; explicit free tier details.

Example prompt costs are provided: simple UI changes cost 0.5–0.9 credits, adding authentication costs 1.2 credits, and building a landing page with images costs 1.7 credits. Credit consumption varies by task complexity and plan.

Official source; specific credit costs for actions are listed.

Lovable offers a free plan with a daily grant of 5 build credits (up to 30 per month), 20 monthly Cloud credits, and 4 AI feature credits. Paid plans add monthly credit balances, daily build grants, and monthly Cloud credits. All plans support unlimited members and are priced by credits, not seats.

Official source; clear plan structure and credit system described.

Bolt offers users the ability to start building for free, as indicated by the prompt to 'start building for free.' No specific details about limits or features of the free tier are provided.

No pricing numbers or detailed plan breakdowns are present in the extracted text.

The Free plan includes 300,000 tokens daily and 1 million tokens monthly, with Bolt branding and a 10MB file upload limit. Website hosting and unlimited databases are included.

Token limits and feature restrictions are explicitly stated.

Bolt now offers four plans: Free ($0), Pro ($25/month), Teams ($30/month per member), and Enterprise (custom pricing). Each plan has different token limits, features, and support levels. Teams plan is billed per member and includes centralized billing and admin controls.

Pricing and features are clearly listed on the official vendor page.

Cognition launched SWE-1.7, its most capable model to date, offering frontier-level intelligence at a much lower cost.

Model launch is clearly stated with cost-performance improvement.

Devin Fusion, a hybrid-model harness, delivers frontier-level coding intelligence at 35% lower cost by using sidekick agents and dynamic routing.

The 35% cost reduction is explicitly stated for the Fusion product.

Cognition introduced the AI Productivity Guarantee for enterprise customers: if Devin delivers less engineering value than paid for, Cognition will fund usage until it does, up to $10M.

Guarantee is capped at $10M and targets enterprise customers.

Most Claude Code surfaces require a Claude subscription or Anthropic Console account. The desktop app specifically requires a paid subscription.

No specific pricing numbers are mentioned, but the requirement for a paid subscription is explicit for the desktop app.

Claude Code is included in the Pro plan for $17/month (annual, $200 billed up front) or $20/month (monthly). The Max plan includes Claude Code for $100/month (Max 5x) and $200/month (Max 20x). Usage limits apply and prices exclude tax.

Pricing details are clearly listed on the official Anthropic Claude Code page.

Replit offers four plans: Starter (Free), Core ($25/month or $20/month billed annually), Pro ($100/month or $95/month billed annually), and Enterprise (custom pricing). Each plan includes different features, such as agent credits, collaborators, and advanced controls. Annual billing provides up to 20% off for Core and 5% off for Pro.

Extracted directly from the official Replit pricing page. Includes clear plan names, prices, and feature differences.

Replit Core is priced at $25 per month, or $20 per month when billed annually (20% discount). The plan includes $25 of monthly credits, up to 5 collaborators, and additional features over the free Starter plan.

Pricing and features are clearly listed on the official Replit pricing page.

Claude Opus 4.7 (fast mode) is now available in Windsurf, offering the full intelligence of Opus 4.7 with approximately 2.5x higher output speeds.

No pricing changes are mentioned, only the availability and performance improvement.

Windsurf has been rebranded and is now called Devin Desktop. The new version is described as a full IDE with an Agent Command Center for managing local and cloud agents from a single interface.

No pricing or feature tier changes are mentioned, only the rebranding and new capabilities.

Codeium offers three individual plans: Free ($0), Pro ($20/month), and Max ($200/month). The Pro plan includes increased quotas, access to OpenAI, Claude, Gemini models, and cloud agents. The Max plan offers significantly higher quotas. Team plans start at $80/month plus $40/month per full dev seat. Enterprise pricing is custom.

Pricing and plan details are clearly listed on the official Codeium pricing page.

Cursor added Google Workspace plugins, allowing coding agents to access Gmail, Google Drive, Calendar, Docs, and Sheets directly from Cursor. Users can now manage files, emails, calendars, and documents without leaving the platform.

Official changelog; feature addition, not a pricing or plan change.

Cursor offers multiple plans: Hobby (Free), Individual ($20/mo), Teams ($40/user/mo), and Enterprise (custom pricing). Each plan includes different feature sets and usage limits, with higher tiers offering increased limits, team features, and advanced controls.

Pricing and plan details are directly from the official Cursor pricing page.

Cursor for iPad is now available on all paid plans, expanding device support for Cursor Pro users. The update includes a new inbox, enhanced review experience, and improved multi-PR session handling.

Official source; device/platform expansion for paid plans.

Cursor introduced 'Cursor Start', a new monthly plan for developers in India, priced at ₹649 per month (tax inclusive), billed in INR via UPI or card. The plan includes access to Grok 4.5 and Composer models, always-on cloud agents, iOS remote control, and plugin support. Available from July 28, 2026.

Official source; clear pricing and plan details for Indian market.

Cursor offers a Hobby (Free) plan, Individual Pro plans starting at $20/month, and Teams plans at $40/user/month. Higher tiers (Pro+, Ultra, Premium, Enterprise) provide increased usage limits and additional features such as priority support, pooled usage, and advanced security.

Pricing and plan details are clearly listed on the official Cursor pricing page.

OpenAI models are now generally available on Amazon Bedrock, expanding the choice of foundation models for customers. This allows users to select from a broader range of models for their generative AI applications.

No specific pricing or plan changes mentioned; announcement is about model availability.

Inference Providers now includes a generous free tier, with additional credits available for PRO users and Team & Enterprise organizations. No extra markup is applied on provider rates.

No specific numbers or limits for the free tier are mentioned in the provided text.

Fireworks AI is moving to prepaid billing starting July 1st. This represents a change from their previous billing model.

No specific pricing numbers mentioned, but the billing model change is explicit and dated.

Serverless inference is priced per token, with example rates for embeddings: up to 150M parameters at $0.008 per 1M input tokens, 150M-350M at $0.016, and Qwen3 8B at $0.1. Training (fine-tuning) is priced per 1M training tokens, ranging from $0.50 to $40.00 depending on model size and method. On-demand GPU pricing is $7.00/hour for H100 and H200, $10.00/hour for B200, and $12.00/hour for B300.

Pricing details are current as per official vendor page. No explicit mention of a change date.

Kimi K3, the first open 3T-class model, is now available via the Together AI API.

Model launch; no pricing details provided.

MiniMax-M3 is now available for efficient inference on Together GPU Clusters.

Model launch; no pricing details provided.

Together AI and Y Combinator announced a partnership to launch the first dedicated GPU cluster for the YC community, eliminating the need for two-year compute contracts.

No explicit pricing or plan changes, but mentions a new offering for YC startups.

Together AI announced an $800M Series C funding round to accelerate the shift to open-source AI. No specific pricing or plan changes were mentioned.

No pricing or plan changes found in the extracted text.

The Together API lists per-1M token prices for various models, such as DeepSeek V4 Pro at $1.74 input/$3.48 output, MiniMax M3 at $0.30 input/$1.20 output, and Kimi K3 at $3.00 input/$15.00 output. Some models have cached output prices, e.g., DeepSeek V4 Pro at $0.20 per 1M tokens (cached).

Pricing is model-specific and includes both input and output token rates. No explicit mention of a change from previous pricing.

A blog post dated October 16, 2025, mentions 'GPT‑OSS Improvements: Prompt Caching & Lower Pricing,' indicating that GroqCloud has reduced its pricing for certain services or models.

No specific price numbers are provided in the extracted text.

Batch API processing is available, offering asynchronous processing of large workloads at 50% lower cost compared to standard rates, with a processing window of 24 hours to 7 days.

Batch API is designed for large-scale asynchronous workloads.

Groq API pricing is listed per million tokens for various models: GPT OSS 20B at $0.075 input/$0.30 output, GPT OSS 120B at $0.15 input/$0.60 output, Llama 3.3 70B at $0.59 input/$0.79 output, Llama 3.1 8B at $0.05 input/$0.08 output, and Qwen 3.6 27B at $0.60 input/$3.00 output. Prompt caching offers a 50% discount on input token prices for cache hits.

Pricing is linear and predictable with no hidden costs. Prompt caching discounts are clearly specified.

Command R received a substantial update in August 2024, now designated as Command R 08-2024 and accessible via model ID command-r-08-2024.

The update includes feature improvements and a new time-stamped model designation.

Command R is priced at $0.15 per 1M input tokens and $0.6 per 1M output tokens, as specified in the official documentation.

Pricing is clearly listed in the official documentation.

No specific pricing or plan details for Command R are listed. The page mentions Command R as a new high-performance model for agentic, multimodal, multilingual AI, but does not provide pricing or plan changes.

Command R is mentioned as a new model, but no pricing or plan information is provided in the extracted text.

Cohere released an updated version of Command R+ in August 2024, designated as command-r-plus-08-2024, with feature improvements for tool use and RAG.

Model versioning now uses time stamps; feature improvements included.

Command R+ is priced at $2.5 per 1M input tokens and $10 per 1M output tokens, as specified in the official documentation.

Pricing is clearly stated in the official documentation.

Trial API keys are automatically created for new accounts and allow free API calls, but are rate limited and not permitted for production or commercial use.

Trial API keys are available in the dashboard upon account creation.

Model Vault pricing is listed per instance: Embed 4 Small is $4.00 hourly or $2,500 monthly, Embed 4 Medium is $5.00 hourly or $3,250 monthly, Rerank 3.5 Medium, Rerank 4 Fast Medium, and Rerank 4 Pro Medium are each $5.00 hourly or $3,250 monthly, and Rerank 4 Pro Large is $10.00 hourly or $6,500 monthly.

Pricing is for fully managed, dedicated model deployments; both hourly and monthly rates are available.

Grok-4.5 model is priced at $2.00 per 1M input tokens and $6.00 per 1M output tokens. No explicit pricing for Grok-3 is mentioned, but this may indicate a pricing update or replacement.

No explicit mention of Grok-3 pricing; Grok-4.5 is now the flagship model.

Speech to Speech starts at $0.05 per minute, Text to Speech is $15.00 per 1M characters, and Speech to Text is $0.10 per hour (REST) or $0.20 per hour (Streaming).

Official source; all voice pricing clearly listed.

Video generation is priced at $0.01 per second, $0.002 per image at 480p, $0.05 per second at 720p, and $0.07 per second.

Official source; detailed per-second and per-image pricing.

Image generation is priced at $0.002 per image for 1K resolution, $0.02 per image for 2K, and $0.02 per image.

Official source; explicit per-image pricing.

Image generation is priced at $0.01 per image for 1K resolution, $0.05 per image for 2K, and $0.07 per image.

Official source; clear per-image pricing.

Grok-4.5 is priced at $2.00 per million input tokens and $6.00 per million output tokens.

Official source; explicit pricing details provided.

Grok-4.5 model is priced at $2.00 per 1M input tokens and $6.00 per 1M output tokens. No explicit pricing for Grok-2 is mentioned in the provided text.

No Grok-2 pricing or plan changes are mentioned; only Grok-4.5 pricing is specified.

Voice API pricing is set at $0.05/min for speech-to-speech, $15.00 per 1M characters for text-to-speech, and $0.10/hr (REST) or $0.20/hr (Streaming) for speech-to-text.

Official pricing for all voice modalities is provided.

Image and video generation pricing is specified: grok-imagine-image-quality at $0.01/img (1K), $0.05/img (2K), $0.07/img; grok-imagine-image at $0.002/img (1K), $0.02/img (2K); grok-imagine-video at $0.01/sec, $0.002/img (480p), $0.05/sec (720p), $0.07/sec.

Multiple pricing tiers for image and video generation are listed.

Grok-4.5 is introduced as xAI's newest flagship model, featuring a 500k context window and advanced capabilities in coding, non-hallucination rate, and agentic tool calling. Pricing is $2.00 per million input tokens and $6.00 per million output tokens.

Pricing and model details are clearly listed on the official xAI API page.

DeepSeek API pricing is $0.0028 per 1M input tokens (cache hit) and $0.14 per 1M input tokens (cache miss) for deepseek-v4-flash; $0.003625 (cache hit) and $0.435 (cache miss) for deepseek-v4-pro. Output tokens are $0.28 (flash) and $0.87 (pro) per 1M. Peak hour pricing will soon be 2x regular prices during specified hours.

Peak hours are 9:00–12:00 and 14:00–18:00 Beijing Time daily; effective date for peak pricing not yet announced.

The deepseek-v4-flash model has been updated to DeepSeek-V4-Flash-0731. The calling method remains unchanged; users can continue to use 'deepseek-v4-flash' to access the latest version.

No pricing or plan changes mentioned; only a model update.

The DeepSeek API service will soon adopt a peak/off-peak pricing policy, with peak hour prices set at 2x the regular rate. The effective date will be announced later.

Applies to all billing items; effective date pending official announcement.

The Responses API currently only supports the DeepSeek-V4-Flash model. Support for DeepSeek-V4-Pro will be added in early August 2026.

Pricing for DeepSeek-V4-Pro is $0.003625 per 1M input tokens (cache hit), $0.435 per 1M input tokens (cache miss), and $0.87 per 1M output tokens. Peak hour pricing will be 2x regular prices during specified hours, effective date to be announced.

Peak hours are 9:00–12:00 and 14:00–18:00 (Beijing Time, UTC+8) daily.

Pricing for DeepSeek-V4-Flash is $0.0028 per 1M input tokens (cache hit), $0.14 per 1M input tokens (cache miss), and $0.28 per 1M output tokens. Peak hour pricing will be 2x regular prices during specified hours, effective date to be announced.

Peak hours are 9:00–12:00 and 14:00–18:00 (Beijing Time, UTC+8) daily.

Mistral AI announced the launch of Mistral Small 4 on March 16, 2026. No pricing or plan details are mentioned in the provided text.

No pricing or plan changes are specified in the extracted text.

Multiple previous versions of Mistral Small (e.g., 3.2, 3.1, 3.0, 2.0) are scheduled for API deprecation and retirement between November 2025 and July 2026, with Mistral Small 4 as the recommended alternative.

Deprecation and retirement dates are clearly listed for each version.

Mistral Small 4 is listed as a new hybrid model unifying instruct, reasoning, and coding in a single efficient model, with Apache 2.0 license. It is positioned as the successor to previous Mistral Small versions, which are scheduled for deprecation and retirement.

Official documentation lists Mistral Small 4 as the current model, with older versions scheduled for deprecation.

Mistral Large 3 is listed as a state-of-the-art, open-weight, general-purpose multimodal model, version 25.12, under the Apache 2.0 license. This appears to be a new or updated model offering.

No pricing or plan details are mentioned; only model listing and version/license information.

Vertex AI has been rebranded as Gemini Enterprise Agent Platform. New customers receive up to $300 in free credits to try Agent Platform and other Google Cloud products.

Official Google Cloud page; clear mention of rebranding and free credit offer.

Custom training pricing is based on selected machine types and accelerator usage, with additional management fees. Spot VMs and Compute Engine reservations are supported and billed per Compute Engine pricing.

Official Google Cloud source; pricing depends on configuration.

ARIMA+ prediction is $5.00 per 1,000 counts. Training is $250.00 per TB multiplied by the number of candidate models and backtesting windows.

Official Google Cloud source; clear pricing for ARIMA+.

Forecasting prediction is $0.20 per 1,000 counts up to 1M, $0.10 per 1,000 counts for 1M–50M, and $0.02 per 1,000 counts above 50M per month. Training is $21.252/hour.

Official Google Cloud source; tiered pricing structure specified.

AutoML model training is $3.465/hour for image data, $18.00/hour for Edge on-device models, and $21.252/hour for tabular data. Deployment and prediction for image data is $1.375/hour (classification) and $2.002/hour (object detection). Batch prediction is $2.222/hour for image data.

Official Google Cloud source; clear hourly rates provided.

Agent Platform pricing remains the same as legacy AI Platform and AutoML products, except that lower-cost, lower-performance machine types and scale-to-zero are no longer supported. New features include optimized TensorFlow runtime, co-hosting models, and no minimum usage duration (charged in 30-second increments).

Official Google Cloud source; specific plan and feature changes detailed.

Gemini 1.5 Flash is listed among the available models in the Gemini API documentation, indicating its official availability for developers.

No pricing or plan changes are mentioned in the provided text.

The documentation lists Gemini 1.5 Pro as one of the available models through the Gemini API, but does not mention any pricing, plan, or feature changes for this model.

No pricing or plan changes are described in the provided text.

Claude Opus 5 is priced at $5 per input million tokens (MTok) and $25 per output MTok. This pricing is listed alongside other Claude models for comparison.

Pricing is from the official Anthropic documentation and is current as of the extracted text.

Claude Haiku 4.5 is priced at $1 per input MTok and $5 per output MTok.

Claude Sonnet 5 is priced at $3 per input MTok and $15 per output MTok. Introductory pricing of $2 per input MTok and $10 per output MTok applies through August 31, 2026.

Introductory pricing valid through August 31, 2026.

Claude Opus 5 is priced at $5 per input MTok and $25 per output MTok.

Claude Fable 5 is priced at $10 per input MTok and $50 per output MTok.

Claude Mythos 5 is available in limited release to approved customers in Project Glasswing starting June 9, 2026. It shares Claude Fable 5's specs and pricing.

Invitation-only; not generally available.

Claude Fable 5 is generally available on the Claude API, Amazon Bedrock, Claude Platform on AWS, Google Cloud, and Microsoft Foundry beginning June 9, 2026.

Claude Sonnet 5 is offered at introductory pricing of $2 per input MTok and $10 per output MTok through August 31, 2026. After this date, standard pricing is $3 per input MTok and $15 per output MTok.

Pricing details are from the official Anthropic documentation.

o4-mini is priced at $1.10 per 1M input tokens and $4.40 per 1M output tokens via API. Effective pricing may be 60–80% lower with prompt caching, depending on repeated context.

Official pricing numbers and caching discount range are provided.

OpenAI launched o4-mini, a compact reasoning model in the o-series, on April 16, 2025. The model is optimized for fast, cost-efficient performance and supports multimodal and agentic capabilities.

Release date and model details are explicitly stated.

OpenAI's o3 model is priced at $2 per 1M input tokens and $8 per 1M output tokens via OpenRouter. The model supports a 200K context window and was released on April 16, 2025.

Pricing is sourced directly from the official OpenRouter page for OpenAI o3.

The o1 model is priced at $15 per 1M input tokens and $60 per 1M output tokens via API.

Official pricing details are clearly stated.

OpenAI launched the o1 model series on December 17, 2024. The model is optimized for STEM tasks and features a 200K context window.

Release date and context window are explicitly mentioned.

GPT-4o mini is priced at $0.15 per 1M input tokens and $0.60 per 1M output tokens, with a context window of 128K tokens.

Official source; explicit pricing and context window stated.

OpenAI has launched GPT-4o mini, a new small model supporting both text and image inputs with text outputs. It is over 60% cheaper than GPT-3.5 Turbo and was released on July 18, 2024.

Official source; clear launch and pricing comparison details provided.

GPT-4o is priced at $2.50 per 1M input tokens and $10 per 1M output tokens. The model is described as being 50% more cost-effective than GPT-4 Turbo.

Pricing is directly stated on the official OpenRouter page for OpenAI GPT-4o.

North Mini Code is listed with a price of $0.00 per million tokens, suggesting it is currently free to use according to the benchmark comparison.

Source is a third-party benchmark site; vendor confirmation required.

Devstral 2 is listed with a price of $0.00 per million tokens, indicating it is currently free to use according to the benchmark comparison.

Source is a third-party benchmark site; vendor confirmation required.

Tool usage with the Agent API is billed per invocation: web_search at $0.0025, fetch_url at $0.00025, people_search and finance_search at $0.005 per invocation or $5 per 1,000 invocations, and sandbox at $0.03 per session (20-minute billing window). Search API is $5 per 1,000 requests. Sonar models have token-based pricing and request fees that vary by context size.

Pricing details are from the official Perplexity documentation and include specific per-invocation and per-token rates.

Anthropic has launched 'Claude for Teachers,' a new offering specifically for educators. No pricing details are mentioned in the provided text.

No pricing or feature specifics provided in the extracted text.

Anthropic has introduced Claude Sonnet 5, which delivers frontier performance across coding, agents, and professional work at scale.

Anthropic has launched Claude Opus 5, described as a step change improvement for the Opus tier, enhancing long-running agents and delivering improvements in coding and professional work.

Gemini 3.6 Flash is now available in Google AI Studio, described as the most intelligent model built for speed, with pricing details provided for various usage tiers.

Model is accessible via the Gemini API with published pricing.

Gemini API now offers three main plans: Free (limited access), Paid (higher limits and advanced features), and Enterprise (custom security, support, and discounts). Each plan provides different access levels and features.

Enterprise plan includes volume discounts and advanced support.

Google has published detailed pricing for the Gemini API, including a free tier with limited access and paid tiers with specific per-1M token rates: $1.50 input and $7.50 output for Standard, $0.75 input and $3.75 output for Batch and Flex, and $2.70 input and $13.50 output for Priority. Additional charges apply for context caching and grounding with Google Search or Maps.

Pricing includes free and paid tiers, with explicit per-token and feature costs.

Anthropic now offers three main plans: Free, Pro, and Max (with 5x and 20x usage options). The Pro plan includes more usage and additional features, while Max provides higher output limits and early access to advanced features.

Plan structure and feature differences are clearly listed on the official site.

Anthropic offers a Free plan at $0, a Pro plan at $17 per month with annual billing ($200 billed up front) or $20 if billed monthly, and a Max plan starting from $100 per month. The Max plan allows users to choose 5x or 20x more usage than Pro. Prices exclude tax and are subject to change.

Pricing details are directly from the official Anthropic pricing page.

Claude Code is included in the Pro plan for $17/month (annual, $200 billed up front) or $20/month (monthly), Max 5x in the Max plan for $100/month, and Max 20x in the Max plan for $200/month. Usage limits apply. Prices exclude tax and are subject to change.

Pricing and plan details are clearly listed on the official Anthropic site.

Cursor for iPad is now available on all paid plans, offering a rebuilt layout for larger screens, sidebar chats, split screen, and enhanced review features. The update also adds an inbox and multi-PR session support on iOS and iPad.

Official changelog; feature expansion for paid users.

Custom training allows use of Spot VMs and Compute Engine reservations, with management fees in addition to infrastructure usage. Pricing is based on selected machine types and attached accelerators, billed separately.

Official Google Cloud source; details on billing and plan options.

Gemini 1.5 Flash is listed among the available models in the Gemini API documentation, indicating its official availability for developers.

No specific pricing or plan changes are mentioned in the provided text.

Pricing for Claude Haiku 4.5 is $1 per input million tokens and $5 per output million tokens.

Pricing for Claude Sonnet 5 is $3 per input million tokens and $15 per output million tokens. Introductory pricing of $2/$10 per MTok applies through August 31, 2026.

Pricing for Claude Opus 5 is $5 per input million tokens and $25 per output million tokens.

Pricing for Claude Fable 5 is $10 per input million tokens and $50 per output million tokens.

Claude Mythos 5 is available in limited release to approved customers in Project Glasswing, starting June 9, 2026. It shares Claude Fable 5's specs and pricing.

Claude Fable 5 is generally available on the Claude API, Amazon Bedrock, Claude Platform on AWS, Google Cloud, and Microsoft Foundry beginning June 9, 2026.

Scenario inputs are now available for all users as of September 23, 2024. The text does not specify if this was previously restricted to certain plans.

No pricing details; only feature availability.

The Starter plan is free and includes daily Agent credits, a built-in database, ability to create slides, videos, animations, publish up to 1 project, and private or password-protected deployments.

Free tier features are explicitly listed on the official pricing page.

Claude Opus 4.7 (fast mode) is now available in Windsurf, offering the full intelligence of Opus 4.7 with approximately 2.5x higher output speeds.

No pricing information provided; model launch only.

Trial API keys are automatically created for new accounts and allow free API calls, but are rate limited and not for production or commercial use.

Production API keys are pay-as-you-go and require owner privileges to obtain.

The DeepSeek API service will soon adopt a peak/off-peak pricing policy, with peak hour prices set at 2x the regular rates. The effective date will be announced later.

Peak hours defined as 9:00–12:00 and 14:00–18:00 Beijing Time daily.

The DeepSeek API service will soon adopt a peak/off-peak pricing policy, with peak hour prices set at 2x the regular rates. The effective date will be announced later.

Peak hours defined as 9:00–12:00 and 14:00–18:00 Beijing Time daily.

OpenAI launched the o3 model on April 16, 2025, with a context window of 200K tokens and a knowledge cutoff of June 2024. Pricing is $2 per 1M input tokens and $8 per 1M output tokens.

Pricing and launch date are explicitly stated on the official OpenRouter page.

GPT-4o, OpenAI's latest AI model, was released on May 13, 2024. It supports both text and image inputs with text outputs and offers improved speed, cost-effectiveness, and multimodal capabilities.

Model launch date and capabilities are clearly stated.

On October 18, 2024, Make AI became available for all users. No pricing or plan details are specified in the text.

No pricing or plan change is mentioned, only feature availability.

Enterprise plan offers all Pro features plus custom seat limits, SSO/SAML, advanced privacy controls, dedicated support, and other enterprise-grade options. Pricing is custom.

Enterprise features and custom pricing are clearly described on the official pricing page.

Replit Pro now includes up to 10 agents working in parallel, database rollbacks for up to 28 days, and premium support.

Feature details are explicitly listed on the official pricing page.

Replit Core now includes unlimited workspaces, up to 2 agents working in parallel, and the ability to remove the 'Made with Replit' badge.

Feature details are explicitly listed on the official pricing page.

Grok-4.5 is introduced as xAI's flagship model for code and general tasks, featuring agentic tool calling, minimal hallucinations, and configurable reasoning. It supports a 500k token context window.

Grok-4.5 is the latest model; Grok-3 is referenced but not detailed.

Anthropic shared the development story of Claude Code, detailing its evolution from an internal CLI to a coding agent.

No pricing, plan, or feature changes; informational news.

Gemini 3.6 Flash is now available in Google AI Studio and via the Gemini API, with specific pricing for input/output tokens and additional features such as context caching and grounding.

Model is positioned as Google's most intelligent model built for speed.

Cursor for iPad is now available on all paid plans, expanding device support for business and enterprise users. The update includes a rebuilt iPad layout, inbox, and enhanced review experience.

Official source; device support expansion for all paid plans.

A blog post dated October 16, 2025, mentions 'GPT‑OSS Improvements: Prompt Caching & Lower Pricing,' indicating that GroqCloud has reduced its pricing for certain services or models.

No specific numbers or detailed pricing changes are provided in the extracted text.

Batch API processing is available, offering asynchronous processing of large workloads at 50% lower cost compared to standard rates, with a processing window of 24 hours to 7 days.

Batch API is designed for large-scale asynchronous workloads.

DeepSeek API will soon implement a peak/off-peak pricing policy, doubling prices during peak hours (9:00–12:00 and 14:00–18:00 Beijing Time). The effective date will be announced later.

No effective date specified yet; applies to all billing items.

Older versions of Mistral Small (3.2, 3.1, 3.0, 2.0) are scheduled for API deprecation and retirement between November 2025 and July 2026, with Mistral Small 4 as the recommended alternative.

Deprecation and retirement dates are specified for each version in the official documentation.

Mistral Small 4, a hybrid model unifying instruct, reasoning, and coding, is now available. It is listed as the successor to previous Mistral Small versions, which are scheduled for deprecation and retirement.

Official documentation lists Mistral Small 4 as the current version, with older versions scheduled for retirement.

Agent Platform pricing remains the same as legacy AI Platform and AutoML products, except lower-cost, lower-performance machine types and scale-to-zero are no longer supported. New pricing includes $3.465/hour for image model training, $1.375-$2.002/hour for deployment and online prediction, $21.252/hour for tabular model training, and tiered forecasting prediction rates starting at $0.20/1,000 counts.

Official Google Cloud pricing page; specific numbers and feature differences are clearly listed.

Cursor for iPad is now available on all paid plans, offering a rebuilt layout for larger screens, sidebar chats, split screen, and enhanced review features. The update also introduces an inbox and improved markup tools.

Official source; feature availability expanded to all paid plans.

Video generation costs $0.01 per second, $0.002 per image at 480p, $0.05 per second at 720p, and $0.07 per second.

Image generation costs $0.002 per image, $0.02 per 1K images, and $0.02 per 2K images.

Image generation costs $0.01 per image, $0.05 per 1K images, and $0.07 per 2K images.

Mistral Large 3 is listed as a new, open-weight, general-purpose multimodal model. It is available under the Apache 2.0 license, version 25.12.

No pricing or plan details are mentioned; only model availability and version.

New customers receive up to $300 in free credits to try the Agent Platform and other Google Cloud products.

Official Google Cloud page; specific free credit amount mentioned.

Vertex AI has been rebranded as Gemini Enterprise Agent Platform, now offering access to Google's latest multimodal models like Gemini 3.5, as well as third-party and open models. The platform provides unified tools for building, scaling, and managing enterprise-grade AI agents.

Official Google Cloud page; clear announcement of rebranding and expanded model access.

Several new features and AI model integrations have been announced, including Claude Fable 5, Claude Opus 5, Claude Sonnet 5, and GPT-5.6 (Luna, Terra, Sol). Feature updates include dark mode, scenario trash, error explanations with AI, and enhanced governance tools.

No pricing or plan changes are mentioned in the extracted text. All changes are general feature or model launches.

Cognition launched SWE-1.7, its most capable model to date, offering frontier-level intelligence at a much lower cost.

Lower cost is mentioned but no specific price is given.

Devin Fusion, a hybrid-model harness, delivers frontier-level coding intelligence at 35% lower cost by using sidekick agents and dynamic routing.

The 35% cost reduction is explicitly stated for the Fusion product.

MiniMax-M3 is now served for efficient inference via Together API, adding to the supported model library.

No pricing or plan changes mentioned.

On-demand B200 GPUs are now available on Together GPU Clusters, expanding the available hardware options.

No pricing details provided.

Enterprise-only models and solutions are available, including models like Minimax M2.7 and batch API processing with 50% lower cost for large-scale asynchronous workloads.

Enterprise access and pricing require contacting Groq for details.

Groq offers a free API key for users to get started, with the option to upgrade as needs grow.

No specific free tier usage limits are detailed in the extracted text.

The DeepSeek API service will soon adopt a peak/off-peak pricing policy, with peak hour prices set at 2x the regular rate. The effective date will be announced later.

Policy applies to all billing items; effective date pending.

Mistral Nemo 12B (v24.07) is scheduled for API deprecation on 5/22/2026 and retirement on 7/31/2026. The suggested alternative is Ministral 3 8B.

Dates and alternative model are explicitly listed in the official documentation.

Mistral AI announced the launch of Mistral Small 4 on March 16, 2026. No pricing or plan details are mentioned in the provided text.

Official vendor source; no pricing or plan changes mentioned.

The documentation lists Gemini 1.5 Pro as one of the available models in the Gemini API, but does not mention any pricing, plan, or feature changes for this model.

No pricing or plan changes are described for Gemini 1.5 Pro in the provided text.

Anthropic introduced 'Claude for Teachers' as a new product offering, but no specific pricing or plan details are provided in the text.

No pricing or plan specifics mentioned.

The Enterprise tier for Gemini API includes all Paid features plus dedicated support, advanced security, provisioned throughput, volume-based discounts, and ML ops/model garden options.

Enterprise tier is for large-scale deployments and requires contacting sales.

Anthropic offers a Free plan at $0, a Pro plan at $17 per month with annual billing ($200 billed up front) or $20 if billed monthly, and a Max plan starting from $100 per month. The Max plan allows users to choose 5x or 20x more usage than Pro.

Pricing and plan details are directly from the official Anthropic pricing page.

Google has introduced three new AI subscription plans: Google AI Plus ($20/mo, 400 GB storage), Google AI Pro ($35/mo, 5 TB storage), and Google AI Ultra (starting at $99/mo, 20 TB+ storage). Each plan offers progressively higher usage limits, storage, and additional features such as YouTube Premium, Google Health Premium, and Google Home Premium.

Pricing and features are clearly listed on the official Google One AI Premium page.

Amazon Nova Pro is now available as part of the Amazon Nova 2 release, offering a highly capable multimodal model with a balance of accuracy, speed, and cost. It supports text, image, and video inputs, and is accessible in multiple regions including US East (N. Virginia), Asia Pacific (Sydney), Europe (London), and AWS GovCloud (US-West).

No specific pricing information is provided in the documentation.

Pipedream has entered into a definitive agreement to be acquired by Workday. The acquisition aims to accelerate the development of AI agent-building products and integrations.

No pricing or plan changes mentioned; this is a general company news item.

Pipedream offers a free account sign-up with no credit card required, and mentions a 'generous free tier' for users. No specific limits or pricing numbers are provided.

No explicit numbers or detailed plan changes are mentioned in the text.

Make has launched several new AI model integrations, including Claude Fable 5, Claude Opus 5, Claude Sonnet 5, and GPT-5.6 (Luna, Terra, and Sol) within its platform.

No pricing or plan changes mentioned; only model availability updates.

Make AI is now available for all users as of October 18, 2024. No pricing or plan details are specified in the provided text.

No mention of pricing or plan changes; only availability.

UiPath offers a Basic plan starting at $25 per month for individuals and small teams, with higher Standard and Enterprise tiers available via contact for pricing. The Basic plan includes personal automations, while Standard and Enterprise add features like enterprise automations, enhanced governance, and self-healing UI automation.

Pricing for Standard and Enterprise plans is not listed and requires contacting sales.

Enterprise plan offers custom workflow executions, unlimited shared projects, 200+ concurrent executions, 365 days of insights, and dedicated support with SLA. Hosted or self-hosted options are available.

Enterprise plan includes extended data retention and external secret store integration.

A free trial is available for Starter and Pro plans with 800 AI Assistant credits and 2,500 executions; no credit card required. Business plan trial requires a credit card and offers 14 days of full access.

Starter and Pro trials have specific resource limits; Business trial is 14 days.

The Starter and Pro plans are hosted by n8n, while the Business plan is self-hosted. Enterprise can be hosted or self-hosted. All plans now include unlimited users and workflows, with shared project limits varying by plan.

Shared projects: Starter (1), Pro (3), Business (6), Enterprise (unlimited).

n8n offers four main plans: Starter at 20€/mo (2,500 executions, 2,300 AI credits), Pro at 50€/mo (10,000 executions, up to 13,700 AI credits), Business at 667€/mo (40,000 executions), and Enterprise with custom pricing. All plans include unlimited users and workflows, with pricing based on monthly workflow executions.

Pricing is billed annually; AI credits and execution limits are specified per plan.

Zapier now uses a unified task-based pricing model for AI steps, code, and SDK across all usage. The Free plan includes 100 tasks per month at $0/month, and paid plans start from 750 tasks per month. Users can pay monthly or yearly (with a 33% discount for yearly payments).

Official Zapier pricing page; specific numbers and plan details provided.

The Roo Code Extension was shut down on May 15th, 2026. The product is no longer available, and users are directed to alternatives. Paying users with billing questions are advised to contact billing@roocode.com.

Official repository notice; no mention of refunds or specific plan details.

Cline is listed as 'Free' on the Visual Studio Marketplace, indicating that the extension does not require payment for installation or use via this channel.

No mention of paid plans or pricing changes; only the free status is indicated.

The continuedev/continue repository is no longer actively maintained and is now read-only for all users. A final 2.0.0 release was made, which included removing anonymous telemetry, pulling out authentication, and bug fixes.

No pricing or plan changes are mentioned; this is a project status update.

Continue has been acquired by Cursor. The open-source codebase remains freely available, and the team expresses gratitude to the community.

No explicit pricing or plan changes mentioned; only acquisition and open-source status confirmed.

Enterprise plans offer volume-based credit pricing for large organizations, with the same unlimited member support as other plans.

Enterprise plan is mentioned as suitable for large organizations.

The free plan now includes a daily grant of 5 build credits (up to 30 per month), 20 monthly Cloud credits, and 4 AI feature credits for trying built-in AI features before subscribing.

Specific free tier credit amounts and AI feature trial credits are detailed.

Example prompt costs are provided: simple UI changes cost 0.5–0.9 credits, adding authentication costs 1.2 credits, and building a landing page with images costs 1.7 credits. In Plan Mode, each message costs 1 credit.

Credit consumption varies by task complexity in Default Mode; Plan Mode is fixed per message.

Lovable offers a free plan with a daily grant of 5 build credits (up to 30 per month), 20 monthly Cloud credits, and 4 AI feature credits. Paid plans include monthly credit balances, daily grants of 5 build credits, and 20 monthly Cloud credits. Workspaces support unlimited members, and plans are priced by credits, not by seats.

No per-seat pricing; all members share workspace credits. Free and paid plan details specified.

Bolt offers users the ability to start building for free, as indicated by the prompt to 'start building for free.' No specific limits or pricing details are provided in the extracted text.

No detailed pricing or plan limits are mentioned; only the existence of a free start is clear.

The Free plan includes 300,000 tokens daily and 1 million tokens monthly, with Bolt branding and a 10MB file upload limit. Pro and higher plans remove branding and increase limits.

Token limits and feature restrictions are explicitly stated.

Bolt now offers four plans: Free ($0), Pro ($25/month), Teams ($30/month per member), and Enterprise (custom pricing). Each plan has different token limits, features, and support levels. Teams plan is billed per member and includes centralized billing and admin controls.

Pricing and features are clearly listed on the official vendor page.

Devin Fusion was introduced, offering frontier-level coding intelligence at 35% lower cost using hybrid-model architecture.

35% cost reduction is explicitly stated.

Cognition launched SWE-1.7, described as their most capable model so far, offering frontier-level intelligence at a much lower cost.

No specific price or percentage given, but clear claim of lower cost.

Cognition introduced the AI Productivity Guarantee for enterprise customers: if Devin delivers less engineering value than paid for, Cognition will fund usage until it does, up to $10M.

Guarantee is specifically for enterprise customers and is capped at $10M.

Most Claude Code surfaces require a Claude subscription or Anthropic Console account. The desktop app specifically requires a paid subscription.

No specific pricing numbers are mentioned, but the requirement for a paid subscription is explicit.

Claude Code is included in the Pro plan for $17 per month with annual billing ($200 billed up front) or $20 if billed monthly. The Max plan includes Claude Code for $100 per month (Max 5x) and $200 per month (Max 20x). Usage limits apply and prices exclude tax.

Pricing details are clearly listed on the official Anthropic site.

Replit offers four plans: Starter (Free), Core ($25/month or $20/month billed annually), Pro ($100/month or $95/month billed annually), and Enterprise (custom pricing). Each plan includes different features, such as agent credits, collaborators, and advanced controls. Annual billing provides up to 20% off for Core and 5% off for Pro.

Pricing and feature details are directly from the official Replit pricing page.

Replit Enterprise offers all Pro features plus custom seat limits, SSO/SAML, advanced privacy controls, dedicated support, and other enterprise-grade options. Pricing is custom and requires contacting sales.

Enterprise features and custom pricing are clearly listed on the official Replit pricing page.

Replit Starter is a free plan that includes daily Agent credits, a built-in database, and the ability to publish up to 1 project.

Free tier features are clearly listed on the official Replit pricing page.

Replit Pro is priced at $100 per month, or $95 per month when billed annually (a 5% discount). The plan includes $100 monthly credits, up to 15 collaborators, and access to more powerful models.

Pricing and features are clearly listed on the official Replit pricing page.

Replit Core is priced at $25 per month, or $20 per month when billed annually (a 20% discount). The plan includes $25 of monthly credits, up to 5 collaborators, and additional features over the free Starter plan.

Pricing and features are clearly listed on the official Replit pricing page.

Windsurf has been rebranded as Devin Desktop, now featuring a full IDE and an Agent Command Center for managing local and cloud agents from a single interface.

No pricing changes mentioned; this is a product/plan evolution.

Codeium offers three individual plans: Free ($0), Pro ($20/month), and Max ($200/month). The Pro plan includes increased quotas, access to OpenAI, Claude, Gemini models, and cloud agents. The Max plan offers significantly higher quotas. Team plans start at $80/month plus $40/month per full dev seat. Enterprise pricing is custom.

Pricing and plan details are clearly listed on the official Codeium pricing page.

Cursor for iPad is now available on all paid plans, offering a rebuilt layout for larger screens, sidebar chats, split screen, and enhanced markup tools. The iPhone and iPad apps now include a full PR review surface and an inbox for organization.

Official source; feature expansion for paid users.

Cursor offers multiple plans: Hobby (Free), Individual ($20/mo), Teams ($40/user/mo), and Enterprise (custom pricing). Each plan includes different feature sets and usage limits, with higher tiers offering increased limits, team features, and advanced controls.

Pricing and plan details are directly from the official Cursor pricing page.

Cursor for iPad is now available on all paid plans, expanding device support for Cursor Pro users. New features include an inbox, full PR review experience, and improved markup tools for iPad and iPhone.

Official changelog; device and feature expansion for paid users.

Cursor introduced 'Cursor Start', a new plan for developers in India, priced at ₹649 per month (tax inclusive), billed in INR with UPI or card. The plan includes access to Grok 4.5 and Composer models, always-on cloud agents, iOS remote control, and workflow extensions. Available from July 28, 2026.

Official changelog; clear pricing and plan details for Indian market.

Cursor offers multiple plans: Hobby (Free), Individual ($20/mo), Pro+, Ultra, Teams ($40/user/mo), and Enterprise (custom pricing). Each plan includes different feature sets and usage limits, with higher tiers offering increased limits and additional features such as priority access, team administration, and enterprise controls.

Pricing and plan details are directly from the official Cursor pricing page.

OpenAI models are now generally available on Amazon Bedrock, expanding the choice of foundation models for customers. This allows users to select from a broader range of models for their generative AI applications.

No specific pricing or plan changes mentioned; announcement is about model availability.

Inference Providers now includes a generous free tier, with additional credits available for PRO users and Team & Enterprise organizations. No extra markup is applied on provider rates.

No specific numbers or limits for the free tier are mentioned in the text.

Fireworks AI is moving to prepaid billing starting July 1st. This represents a change from their previous billing method.

Source is official and provides a specific date for the billing change.

Serverless inference is priced per token, with $0.008 per 1M input tokens for models up to 150M parameters, $0.016 for 150M-350M, and $0.1 for Qwen3 8B. Training is priced per 1M training tokens, ranging from $0.50 to $40.00 depending on model size and fine-tuning method. On-demand GPU pricing is $7.00/hour for H100 and H200, $10.00/hour for B200, and $12.00/hour for B300.

Pricing details are current as per official vendor page. Includes both inference and training costs, as well as on-demand GPU rates.

Kimi K3, the first open 3T-class model, is now available via the Together AI API. The blog post includes benchmarking and cost comparisons, but no specific price is mentioned in the extracted text.

Model launch; cost comparisons referenced but no explicit pricing.

MiniMax-M3 is now available for efficient inference on Together AI's platform. No pricing or plan details are provided.

Model launch announcement; no pricing specifics.

Together AI and Y Combinator announced a partnership to launch the first dedicated GPU cluster for the YC community, eliminating the need for two-year compute contracts and providing YC startups with faster access to GPUs.

No explicit pricing or plan changes, but mentions a shift away from long-term contracts.

Together AI announced an $800M Series C funding round to accelerate the shift to open-source AI. The announcement discusses the economics of closed models and outlines future plans, but does not mention specific pricing or plan changes.

No specific pricing or plan changes mentioned in the extracted text.

The Together API lists updated per-1M token prices for various models, such as DeepSeek V4 Pro at $1.74 input/$3.48 output, MiniMax M3 at $0.30 input/$1.20 output, and Kimi K3 at $3.00 input/$15.00 output. Prices are specified for both input and output tokens, with some models offering cached rates.

Pricing details are directly from the official Together AI pricing page. No explicit mention of previous prices or changes, but current rates are clearly listed.

Groq API pricing is listed per million tokens for various models: GPT OSS 20B at $0.075 input/$0.30 output, GPT OSS 120B at $0.15 input/$0.60 output, Llama 3.3 70B at $0.59 input/$0.79 output, Llama 3.1 8B at $0.05 input/$0.08 output, and Qwen 3.6 27B at $0.60 input/$3.00 output. Prompt caching offers a 50% discount on input token prices for cache hits.

Pricing is linear and predictable; prompt caching discounts input token costs by 50%.

Cohere released an updated version of Command R (command-r-08-2024) in August 2024, featuring improved tool use, retrieval augmented generation, and multilingual capabilities.

Model versioning now uses time stamps; this is the latest official release.

Command R is priced at $0.15 per 1M input tokens and $0.6 per 1M output tokens. These rates are specified for the August 2024 release.

Pricing is clearly listed in the official documentation for the August 2024 release.

No specific pricing or plan details for Command R are listed. The page mentions Command R as a new high-performance model for agentic, multimodal, multilingual AI, but does not provide pricing or plan changes.

Command R is mentioned as 'NEW' but no pricing or plan information is provided on the page.

Cohere released an updated version of Command R+ in August 2024, designated as command-r-plus-08-2024, with feature improvements for tool use and RAG.

Model update is timestamped and described in the official documentation.

Command R+ is priced at $2.5 per 1M input tokens and $10 per 1M output tokens, as specified in the official documentation.

Pricing is clearly stated in the official documentation.

Trial API keys are automatically created for new accounts and allow free API calls, but are rate limited and not for production or commercial use.

Production API keys are charged on a pay-as-you-go basis.

Model Vault pricing is listed per instance: Embed 4 Small is $4.00 hourly or $2,500 monthly; Embed 4 Medium is $5.00 hourly or $3,250 monthly; Rerank 3.5 Medium, Rerank 4 Fast Medium, and Rerank 4 Pro Medium are each $5.00 hourly or $3,250 monthly; Rerank 4 Pro Large is $10.00 hourly or $6,500 monthly.

Pricing is for fully managed, dedicated model deployments. Billing can be hourly or monthly.

Grok-4.5 model is priced at $2.00 per 1M input tokens and $6.00 per 1M output tokens. Voice API pricing starts at $0.05 per minute for Agent, $15.00 per 1M characters for TTS, $0.10 per hour for STT (Batch), and $0.20 per hour for STT (Streaming).

Pricing is for Grok-4.5 and related APIs as listed on the official documentation.

Grok-4.5 is priced at $2.00 per million input tokens and $6.00 per million output tokens.

Official source; explicit pricing details.

Grok-4.5 model is priced at $2.00 per 1M input tokens and $6.00 per 1M output tokens. Voice API starts at $0.05 per minute for Agent, $15.00 per 1M characters for TTS, $0.10 per hour for batch STT, and $0.20 per hour for streaming STT. Image generation starts at $0.02 per image, and video generation starts at $0.05 per second.

Pricing is for Grok-4.5 and related APIs; Grok-2 is not explicitly priced in this text.

Voice API pricing is listed: Speech to Speech starts at $0.05/min, Text to Speech at $15.00 per 1M characters, and Speech to Text at $0.10/hr (REST) or $0.20/hr (Streaming).

Official pricing breakdown for all voice-related services.

Image and video generation pricing is specified: grok-imagine-image-quality at $0.01/img (1K), $0.05/img (2K), $0.07/img; grok-imagine-image at $0.002/img (1K), $0.02/img (2K); grok-imagine-video at $0.01/sec, $0.002/img (480p), $0.05/sec (720p), $0.07/sec.

Multiple price points for different resolutions and models are provided.

Grok-4.5 is introduced as xAI's newest flagship model, supporting both non-reasoning and reasoning modes, with a 500k context window. Pricing is $2.00 per million input tokens and $6.00 per million output tokens.

Official pricing and model details are clearly listed on the vendor's API page.

DeepSeek announced that its API service will soon adopt a peak/off-peak pricing policy. During peak hours (9:00–12:00 and 14:00–18:00 Beijing Time), prices will be 2x the regular rates for all billing items. The effective date will be announced officially.

Official source; specific price multipliers and peak hours provided, but effective date pending.

The deepseek-v4-flash model has been updated to DeepSeek-V4-Flash-0731. The calling method remains unchanged; users can continue to use 'deepseek-v4-flash' to access the latest version.

No pricing or plan changes mentioned; only a model update.

The Responses API currently only supports DeepSeek-V4-Flash. Support for DeepSeek-V4-Pro will be added in early August 2026.

Pricing for DeepSeek-V4-Pro is $0.003625 per 1M input tokens (cache hit), $0.435 per 1M input tokens (cache miss), and $0.87 per 1M output tokens. Peak hour pricing will be 2x regular prices during specified hours, with the effective date to be announced.

Peak hours are 9:00–12:00 and 14:00–18:00 Beijing Time daily. Peak pricing not yet in effect.

Pricing for DeepSeek-V4-Flash is $0.0028 per 1M input tokens (cache hit), $0.14 per 1M input tokens (cache miss), and $0.28 per 1M output tokens. Peak hour pricing will be 2x regular prices during specified hours, with the effective date to be announced.

Peak hours are 9:00–12:00 and 14:00–18:00 Beijing Time daily. Peak pricing not yet in effect.

Mistral Nemo 12B (version 24.07) is scheduled for API deprecation on 5/22/2026 and retirement on 7/31/2026. The suggested alternative is Ministral 3 8B.

Source is official and provides specific deprecation and retirement dates.

Mistral AI announced the launch of Mistral Small 4 on March 16, 2026. No pricing or plan details are mentioned in the provided text.

No pricing or plan changes are specified in the extracted text.

Mistral Small 3.2 is scheduled for API deprecation on 4/30/2026 and retirement on 7/31/2026. The recommended alternative is Mistral Small 4.

Clear deprecation and retirement timeline provided.

Mistral Small 4 is listed as a new hybrid model unifying instruct, reasoning, and coding in a single efficient model. It is available under the Apache 2.0 license.

No pricing or plan details are mentioned; only model availability and description.

Mistral Large 3 is listed as a new, state-of-the-art, open-weight, general-purpose multimodal model. It is available under the Apache 2.0 license, version 25.12.

No pricing or plan details are mentioned; only model availability and version.

Vertex AI has been rebranded as Gemini Enterprise Agent Platform. New customers receive up to $300 in free credits to try Agent Platform and other Google Cloud products.

The rebranding and free credit offer are stated clearly on the official Google Cloud page.

ARIMA+ forecasting prediction is $5.00 per 1,000 counts. Training is $250.00 per TB multiplied by the number of candidate models and backtesting windows.

Official Google Cloud source; clear pricing for ARIMA+ forecasting and training.

Forecasting prediction is $0.20 per 1,000 counts up to 1M, $0.10 per 1,000 counts for 1M–50M, and $0.02 per 1,000 counts above 50M per month. Training is $21.252/hour.

Official Google Cloud source; tiered pricing structure for forecasting predictions.

AutoML model training is $3.465/hour for image data, $18.00/hour for Edge on-device models, and $21.252/hour for tabular data. Deployment and prediction for image data is $1.375/hour (classification) and $2.002/hour (object detection). Batch prediction is $2.222/hour.

Official Google Cloud source; clear hourly rates provided for various AutoML operations.

Agent Platform pricing remains the same as legacy AI Platform and AutoML products, except that lower-cost, lower-performance machine types and scale-to-zero are no longer supported. Agent Platform now charges in 30-second increments with no minimum usage duration.

Official Google Cloud source; specific changes to supported machine types and billing increments.

The documentation lists Gemini 1.5 Flash as one of the available models in the Gemini API, but does not mention any pricing, plan, or feature changes for this model.

No pricing or plan changes are specified for Gemini 1.5 Flash in the provided text.

Claude Opus 5 is priced at $5 per input million tokens (MTok) and $25 per output MTok, as listed in the official model comparison table.

Pricing is from the official Anthropic documentation and is current as of the extracted text.

Claude Haiku 4.5 is priced at $1 per input MTok and $5 per output MTok. The model supports a 200k token context window and a maximum output of 64k tokens.

Pricing is from the official Anthropic documentation. No mention of free tier or enterprise-specific changes.

Claude Sonnet 5 is priced at $3 per input MTok and $15 per output MTok. An introductory pricing of $2 per input MTok and $10 per output MTok applies through August 31, 2026.

Pricing is official and includes a time-limited introductory offer.

The API pricing for o4-mini is $1.10 per 1M input tokens and $4.40 per 1M output tokens. Effective pricing may be 60–80% cheaper with prompt caching, depending on repeated context.

Official source with explicit pricing details.

OpenAI has launched o4-mini, a compact reasoning model in the o-series, optimized for fast and cost-efficient performance. The model was released on April 16, 2025, with a context window of 200K tokens.

Source is official and provides clear release date and model details.

OpenAI's o3 model is priced at $2 per 1M input tokens and $8 per 1M output tokens via OpenRouter. The model supports a 200K context window and was released on April 16, 2025.

Pricing is sourced directly from the official OpenRouter page for OpenAI o3.

The o1 model is priced at $15 per 1M input tokens and $60 per 1M output tokens via API.

Official pricing details are clearly stated.

OpenAI launched the o1 model series on December 17, 2024. The model is optimized for STEM tasks and features a 200K context window.

Release date and context window are explicitly mentioned.

GPT-4o mini is priced at $0.15 per 1M input tokens and $0.60 per 1M output tokens, with a context window of 128K tokens.

Official pricing details provided for API usage.

OpenAI has launched GPT-4o mini, a new small model supporting both text and image inputs with text outputs. It is over 60% cheaper than GPT-3.5 Turbo and was released on July 18, 2024.

Official source confirms launch date and relative pricing.

GPT-4o is priced at $2.50 per 1M input tokens and $10 per 1M output tokens. This model is twice as fast and 50% more cost-effective compared to GPT-4 Turbo.

Pricing is directly stated on the official OpenRouter page for OpenAI GPT-4o.

North Mini Code is listed with a price of $0.00 per million tokens, indicating it is among the cheapest models in the comparison.

Source is a third-party benchmark site; official vendor confirmation required.

Devstral 2 is listed with a price of $0.00 per million tokens, making it one of the cheapest models according to the benchmark.

Source is a third-party benchmark site; official vendor confirmation required.

Perplexity Pro's API pricing includes token-based charges for different models (e.g., Sonar Pro: $3 per 1M input tokens, $15 per 1M output tokens) and request fees based on search context size (e.g., $6/$10/$14 per 1,000 requests for low/medium/high context). Tool usage (e.g., web_search: $0.0025 per invocation) and Search API ($5 per 1,000 requests) are also specified.

Pricing details are from the official Perplexity documentation and include specific numbers for various models and tools.

Anthropic announced Claude for Teachers, a new product offering specifically for educators.

No pricing or plan changes mentioned; announcement is for a new product.

Anthropic introduced Claude Sonnet 5, which delivers frontier performance across coding, agents, and professional work at scale.

No pricing or plan changes mentioned; announcement is for a new model.

Anthropic has launched Claude Opus 5, described as a step change improvement for the Opus tier, enhancing long-running agents and delivering improvements in coding and professional work.

No pricing or plan changes mentioned; announcement is for a new model.

The Free tier for Gemini API provides limited access to certain models, free input and output tokens, and Google AI Studio access. Content from the free tier may be used to improve Google products.

Free tier has usage and model restrictions; content may be used for product improvement.

Gemini API now offers three plans: Free (limited access), Paid (higher rate limits, advanced features), and Enterprise (custom security, support, compliance, and volume discounts). Each plan provides different levels of access and features.

Enterprise plan includes dedicated support, advanced security, and volume-based discounts.

Google has published detailed pricing for the Gemini API, including a free tier with limited access and paid tiers with specific per-token and per-request costs. For example, Gemini 3.6 Flash is priced at $1.50 per 1M input tokens and $7.50 per 1M output tokens in the paid tier, with additional charges for context caching and grounding with Google Search or Maps.

Pricing includes free and paid tiers, with explicit costs for input/output tokens, context caching, and grounding features.

Anthropic now offers three main plans: Free ($0), Pro ($17/month with annual billing or $20/month billed monthly), and Max (starting at $100/month with options for 5x or 20x Pro usage). Each plan includes different features and usage limits.

Pricing and plan details are directly from the official Anthropic pricing page.

Custom training pricing is based on selected machine types and accelerator usage, with additional management fees. Spot VMs and Compute Engine reservations are supported and billed per Compute Engine pricing.

No specific hourly rates given in this extract, but pricing structure is clearly described.

Pipedream has entered into a definitive agreement to be acquired by Workday. The announcement states that Pipedream's AI agent-building products and integrations will be combined with Workday's offerings.

No pricing or plan changes are mentioned in the provided text.

Cursor for iPad is now available on all paid plans, offering a rebuilt layout for larger screens, sidebar chats, split screen, and enhanced review and markup features. The update also introduces an inbox and improved PR review experience on iPhone and iPad.

Official changelog; feature expansion to all paid plans.

A blog post dated October 16, 2025, mentions 'GPT‑OSS Improvements: Prompt Caching & Lower Pricing,' indicating that GroqCloud has reduced its pricing for certain services or models.

No specific price numbers are provided in the extracted text.

The Responses API currently only supports the DeepSeek-V4-Flash model. Support for DeepSeek-V4-Pro will be added in early August 2026.

Mistral Small 3.2 will be deprecated on 4/30/2026 and retired on 7/31/2026. The recommended alternative is Mistral Small 4.

Deprecation and retirement schedule is clearly listed in the official documentation.

Speech to Speech starts at $0.05 per minute, Text to Speech is $15.00 per 1M characters, and Speech to Text is $0.10 per hour (REST) or $0.20 per hour (Streaming).

Official pricing for voice API.

The Responses API currently only supports the deepseek-v4-flash model. Support for deepseek-v4-pro will be added in early August 2026.

Cursor Router, an intelligent model router, is now available and on by default for Teams plans. It offers optimization modes (Intelligence, Balance, Cost) and admin controls for routing requests to appropriate models based on cost and quality.

Official source; feature launch with admin controls and optimization modes.

Cline is listed as a free extension on the Visual Studio Marketplace, offering autonomous coding features within VS Code. No pricing or plan changes are mentioned in the provided text.

No explicit pricing or plan changes; extension is currently free.

Devin Fusion, a hybrid-model harness, was introduced, claiming to deliver frontier-level coding intelligence at 35% lower cost using sidekick agents and dynamic routing.

35% cost reduction is explicitly stated.

Cognition launched SWE-1.7, described as their most capable model to date, offering frontier-level intelligence at a much lower cost.

No specific pricing numbers provided, but cost reduction is emphasized.

Video generation is priced at $0.01 per second, $0.002 per image (480p), $0.05 per second (720p), and $0.07 per second.

Image generation is priced at $0.002 per image, $0.02 per 1K images, and $0.02 per 2K images.

Image generation is priced at $0.01 per image, $0.05 per 1K images, and $0.07 per 2K images.

Custom training allows use of Spot VMs and Compute Engine reservations, with management fees in addition to infrastructure usage. Spot VMs are billed per Compute Engine Spot VM pricing.

Official Google Cloud source; describes plan and billing options.

The Make Community page lists recent feature spotlights and app updates, including new AI features, scenario management tools, and integrations with models like Claude and GPT-5.6. No pricing or plan changes are mentioned.

No pricing, plan, or free tier changes found in the extracted text.

The free plan now includes 100 tasks per month and access to Zap workflows, Tables, and Forms, allowing users to start automating with AI at no cost.

Official source; clear mention of free tier features and limits.

Pro plan is priced at $25 per month billed monthly. Teams plan is $30 per month per member billed monthly. Free plan is $0. Enterprise plan is custom pricing.

Official pricing details provided.

Enterprise-only models and solutions are available, including models like Minimax M2.7 and batch API processing for large-scale workloads with 50% lower cost compared to standard rates.

Enterprise access and pricing require contacting Groq directly.

Mistral Nemo 12B (version 24.07) is scheduled for API deprecation on 5/22/2026 and retirement on 7/31/2026. The suggested alternative is Ministral 3 8B.

Source is official and provides specific deprecation and retirement dates.

The DeepSeek API service will soon adopt a peak/off-peak pricing policy, with peak hour prices set at 2x the regular rate. The effective date will be announced later.

Policy applies to all billing items; effective date pending.

Anthropic offers a Free plan at $0, a Pro plan at $17 per month with annual billing ($200 billed up front) or $20 if billed monthly, and a Max plan starting from $100 per month. The Max plan allows users to choose 5x or 20x more usage than Pro.

Pricing and plan details are directly from the official Anthropic pricing page.

Scenario inputs are now available for all users as of September 23, 2024. The text does not specify if this was previously restricted to certain plans.

No pricing details; only feature availability change.

New customers receive up to $300 in free credits to try Gemini Enterprise Agent Platform and other Google Cloud products.

This is a clear, official statement about the free tier for new users.

Vertex AI has been rebranded as Gemini Enterprise Agent Platform, offering a unified platform for building, scaling, and optimizing enterprise-grade AI agents. The platform now includes access to Google's latest multimodal models like Gemini 3.5 and third-party models.

Official Google Cloud page confirms the rebranding and expanded model access.

Replit Enterprise offers all Pro features plus custom seat limits, SSO/SAML, advanced privacy controls, dedicated support, and other enterprise-grade options. Pricing is custom and requires contacting sales.

Enterprise features and custom pricing are clearly listed on the official Replit pricing page.

Replit Starter is a free plan that includes daily Agent credits, a built-in database, and the ability to publish up to 1 project with private or password-protected deployments.

Free tier features are clearly listed on the official Replit pricing page.

Kimi K3, the first open 3T-class model, is now available via the Together AI API.

MiniMax-M3 is now available for efficient inference on Together GPU Clusters.

Forecasting predictions are $0.20 per 1,000 counts up to 1M, $0.10 per 1,000 counts from 1M to 50M, and $0.02 per 1,000 counts above 50M per month. Training is $21.252/hour.

Official tiered pricing for forecasting predictions.

Make AI is now available for all users as of October 18, 2024. No pricing or plan details are specified in the provided text.

No pricing or plan changes mentioned; only availability.

The Free plan includes 300,000 tokens daily and 1 million tokens monthly, with Bolt branding and a 10MB file upload limit. Pro and higher plans remove branding and increase limits.

Token limits and feature differences are explicitly stated.

Claude Opus 4.7 (fast mode) is now available in Windsurf, offering the full intelligence of Opus 4.7 with approximately 2.5x higher output speeds.

Official announcement specifies model availability and performance improvement.

Cursor for iPad is now available on all paid plans, offering a rebuilt layout for larger screens, sidebar chats, split screen, and enhanced review and markup features. The update also introduces an inbox and improved PR review experience on iPhone and iPad.

Official source; feature expansion to iPad for all paid users.

MiniMax-M3 model is now served for efficient inference via Together API.

No pricing or plan changes mentioned.

On-demand B200 GPUs are now available on Together GPU Clusters, expanding hardware options for users.

No pricing details provided.

Batch API processing is available at 50% lower cost compared to standard rates, allowing asynchronous processing of large workloads.

Batch API enables large-scale asynchronous requests with reduced pricing.

Compass, Cohere's intelligent search and discovery system, is available only with custom enterprise pricing.

No public pricing; must contact sales.

North, Cohere's all-in-one AI platform, is available only with custom enterprise pricing.

No public pricing; must contact sales.

Production API keys are available to users with Owner privileges and are charged on a pay-as-you-go basis.

Production keys are not free and require a workflow to obtain.

Mistral AI announced the launch of Mistral Small 4 on March 16, 2026. No pricing or plan details are mentioned in the provided text.

No pricing or plan changes are specified in the extracted text.

Mistral Small 3.2 is scheduled for API deprecation on 4/30/2026 and retirement on 7/31/2026. The recommended alternative is Mistral Small 4.

Clear deprecation and retirement timeline provided.

Mistral Small 4 is listed as a new hybrid model unifying instruct, reasoning, and coding in a single efficient model. It is available under the Apache 2.0 license.

No pricing or plan details are mentioned; only model availability and description.

Mistral Large 3 is listed as a new, open-weight, general-purpose multimodal model. It is available under the Apache 2.0 license, version 25.12.

No pricing or plan details are mentioned in the provided text.

Agent Platform pricing remains the same as legacy AI Platform and AutoML products, except that lower-cost, lower-performance machine types and scale-to-zero are no longer supported. New pricing includes $3.465/hour for image model training, $1.375-$2.002/hour for deployment and online prediction, $21.252/hour for tabular model training, and tiered forecasting prediction rates starting at $0.20/1,000 counts.

Official Google Cloud pricing page; specific numbers and feature differences are listed.

OpenAI's o3 model is priced at $2 per 1M input tokens and $8 per 1M output tokens via OpenRouter. The model supports a 200K context window and was released on April 16, 2025.

Pricing is explicitly stated on the official OpenRouter page for OpenAI o3.

Anthropic announced Claude for Teachers, a new product offering tailored for educators.

No pricing or plan changes mentioned; announcement is for a new product.

Gemini 3.6 Flash is now available in the Gemini API, offering high-speed, intelligent model access with specific pricing for input and output tokens, context caching, and grounding features.

Model is accessible in both free and paid tiers with published rates.

The Gemini API now offers three plans: Free (limited access), Paid (higher rate limits, advanced features), and Enterprise (custom security, support, compliance, and volume discounts). Each plan provides different levels of access and features.

Enterprise plan includes dedicated support, advanced security, and volume-based discounts.

Google has introduced three new AI subscription plans: Google AI Plus ($20/mo, 400 GB storage), Google AI Pro ($35/mo, 5 TB storage), and Google AI Ultra (starting at $99/mo, 20 TB+ storage). Each plan offers progressively higher usage limits, storage, and additional features such as YouTube Premium, Google Health Premium, and Google Home Premium.

Pricing and features are clearly listed on the official Google One AI Premium page.

Amazon Nova Pro is a newly available multimodal foundation model accessible via Amazon Bedrock. It offers a combination of accuracy, speed, and cost for a wide range of tasks, supporting text, image, and video inputs, and is available in multiple regions.

No specific pricing or plan changes are mentioned; this is a model launch with detailed feature specifications.

Pipedream has entered into a definitive agreement to be acquired by Workday. The acquisition aims to accelerate the development of AI agent-building products and integrations.

No pricing or plan changes are mentioned; this is a general company news update.

Pipedream offers a free account sign-up with no credit card required, and mentions a 'generous free tier' for users. No specific limits or pricing numbers are provided.

No explicit details on free tier limits or paid plans in the provided text.

Make has launched several new AI model integrations, including Claude Fable 5, Claude Opus 5, Claude Sonnet 5, and GPT-5.6 (Luna, Terra, and Sol) within its platform. These updates expand the available AI capabilities for users.

No pricing or plan changes mentioned; only model availability updates.

On October 18, 2024, Make AI became available for all users. No pricing or plan details are specified in the text.

No pricing or plan change is mentioned, only feature availability.

UiPath offers a Basic plan for Automation Cloud Test Cloud starting at $25 per month, targeting individuals and small teams. Standard and Enterprise plans are available with additional features, but pricing for these tiers requires contacting sales.

Pricing for Standard and Enterprise plans is not listed; only Basic plan has a specific price.

Enterprise plan includes unlimited shared projects, 200+ concurrent executions, 365 days of insights, external secret store integration, log streaming, extended data retention, and dedicated support with SLA.

Enterprise plan is available as hosted or self-hosted; pricing is custom.

A free trial is available for Starter and Pro plans with 800 AI Assistant credits and access to all Pro features. No credit card is required for Starter and Pro trials; Business plan trial requires a credit card.

Free trial limits: 5 concurrent executions, 320MiB RAM, 10 millicore CPU, 2,500 executions.

All plans now include unlimited users and workflows. Starter, Pro, and Business plans have defined limits for workflow executions and AI credits. Enterprise plan offers custom executions and additional features like extended data retention and dedicated support.

AI Assistant is marked as 'coming soon' for Business and Enterprise plans.

Starter plan is 20€/mo (billed annually) for 2,500 workflow executions and 2,300 AI credits/month. Pro plan is 50€/mo (billed annually) for 10,000 executions and up to 13,700 AI credits/month. Business plan is 667€/mo (billed annually) for 40,000 executions. Enterprise plan is custom pricing. All plans include unlimited users and workflows.

Pricing is based on workflow executions, not per step or user. AI credits included per plan.

Zapier has unified its pricing so that AI steps, code, and SDK now all follow the same task-based pricing model. The Free plan includes 100 tasks per month at $0/month, and paid plans start from $1.

Official Zapier pricing page; clear mention of new unified task-based pricing and free tier details.

The Roo Code Extension was shut down on May 15th, 2026. The product is no longer available, and users are directed to alternatives. Paying users with billing questions are advised to contact billing@roocode.com.

Official repository notice; no mention of refunds or specific plan details.

Cline is listed as 'Free' on the Visual Studio Marketplace, indicating that the extension does not require payment for installation or use as of the current version.

No mention of paid plans or pricing changes; extension is currently free.

The continuedev/continue repository is no longer actively maintained and is now read-only for all users. A final 2.0.0 release was made, which included removing anonymous telemetry, pulling out authentication, and bug fixes.

No pricing or plan changes are mentioned; this is a project status update.

Continue has been acquired by Cursor. The open-source codebase remains freely available, and the team expresses gratitude to the community.

No explicit pricing or plan changes mentioned; acquisition and open-source status confirmed.

The free plan now includes a daily grant of 5 build credits (up to 30 per month), 20 monthly Cloud credits, and 4 AI feature credits for trial use.

Official source; explicit numbers for free tier.

Enterprise plans offer volume-based credit pricing for large organizations. Business plans have higher prices due to additional features, but credit usage for running apps and AI features is the same as Pro.

Official source; enterprise and business plan distinctions described.

Credits are consumed per action, e.g., 0.5 credits to update button styles, 0.9 to remove a footer, 1.2 to add authentication, and 1.7 to build a landing page. Credit consumption varies by task complexity and plan.

Official source; specific credit costs for example actions provided.

Lovable offers a free plan with a daily grant of 5 build credits (up to 30 per month), 20 monthly Cloud credits, and 4 AI feature credits. Paid plans add monthly credit balances, daily build grants, and monthly Cloud credits. Workspaces support unlimited members with pricing based on credits, not seats.

Official source; specific numbers for free plan and credit system described.

Bolt offers users the ability to start building for free, as indicated by the prompt to 'start building for free.' No specific limits or pricing details are provided in the extracted text.

No detailed pricing or plan limits are mentioned; only the existence of a free start is clear.

Bolt now offers four plans: Free ($0), Pro ($25/month), Teams ($30/month per member), and Enterprise (custom pricing). Each plan has different token limits, features, and support levels. Teams plan is billed per member and includes centralized billing and admin controls.

Pricing and features are clearly listed on the official vendor page.

Cognition launched SWE-1.7, its most capable model to date, offering frontier-level intelligence at a much lower cost.

Lower cost is mentioned but no specific price is given.

Devin Fusion, a hybrid-model harness, delivers frontier-level coding intelligence at 35% lower cost by using sidekick agents and dynamic routing.

The 35% cost reduction is explicitly stated for the Fusion product.

Cognition introduced the AI Productivity Guarantee for enterprise customers: if Devin delivers less engineering value than paid for, Cognition will fund usage until it does, up to $10M.

Guarantee is capped at $10M and targets enterprise customers.

Most Claude Code surfaces require a Claude subscription or Anthropic Console account. The desktop app specifically requires a paid subscription.

No specific pricing numbers are mentioned, but the requirement for a paid subscription is explicit.

Claude Code is included in the Pro plan for $17/month (annual, $200 billed up front) or $20/month (monthly). The Max plan includes Claude Code for $100/month (Max 5x) and $200/month (Max 20x). Usage limits apply. Prices exclude tax.

Pricing is clearly listed for Pro and Max plans with specific numbers and billing details.

Replit offers four plans: Starter (Free), Core ($25/month or $20/month billed annually), Pro ($100/month or $95/month billed annually), and Enterprise (custom pricing). Each plan includes different features, such as agent credits, collaborators, and advanced controls.

Pricing and features are clearly listed on the official Replit pricing page.

The Starter plan is free and includes daily Agent credits, a built-in database, ability to create slides, videos, animations, publish up to 1 project, and private or password-protected deployments.

Free tier features are explicitly stated on the official pricing page.

Enterprise plan includes all Pro features plus custom seat limits, SSO/SAML, advanced privacy controls, design system support, data warehouse connections, custom groups, dedicated support, single-tenant environments, region selection, static outbound IPs, and VPC peering. Pricing is custom.

Enterprise features and custom pricing are clearly listed on the official pricing page.

Replit Pro now offers access to the most powerful models, database rollbacks for up to 28 days, and premium support, in addition to all Core features.

Feature list is explicitly stated on the official pricing page.