AI Model Tracker

All major AI models at a glance, grouped by provider. Pricing, context window, capabilities and release dates.

30 models Updated weekly Prices and specs
OpenAI logo

OpenAI

6 of 13 models
  • GPT-5.5 Current
    Vision Thinking Tool use Computer Use Cache
    Context
    1.000K
    Input/1M
    —
    Output/1M
    —
    Release
    Apr 2026

    Latest OpenAI flagship, 6 weeks after GPT-5.4. Fewer tokens per task, same latency. API not yet available, ChatGPT and Codex only. Pro variant available.

  • GPT-5.4 pro Current
    Vision Thinking Tool use Cache
    Context
    1.050K
    Input/1M
    $30.00
    Output/1M
    $180.00
    Release
    —

    Premium GPT-5.4 variant for the most demanding tasks. Source: openai.com/api/pricing.

  • GPT-5.4 mini Current
    Vision Thinking Tool use Cache
    Context
    400K
    Input/1M
    $0.75
    Output/1M
    $4.50
    Release
    —

    Compact GPT-5.4 variant, competitively priced. Source: openai.com/api/pricing.

  • GPT-5.4 nano Current
    Vision Tool use
    Context
    400K
    Input/1M
    $0.20
    Output/1M
    $1.25
    Release
    —

    Smallest GPT-5.4 variant, for high-volume and simple tasks. Source: openai.com/api/pricing.

  • o4-mini Current
    Reasoning Code Wiskunde Tool use
    Context
    200K
    Input/1M
    $1.10
    Output/1M
    $4.40
    Release
    Jun 2025

    Compact reasoning model from the o-series. Strong price-to-performance for AIME and coding. Source: openai.com/api/pricing.

  • GPT-5.3 Codex Current
    Code Agent workflow Tool use
    Context
    400K
    Input/1M
    $1.75
    Output/1M
    $14.00
    Release
    Jan 2026

    Coding-focused variant from the GPT-5.3 line. Same pricing as gpt-5.2. Source: openrouter.ai.

Anthropic logo

Anthropic

6 models
  • Claude Fable 5 Current
    Context
    1.000K
    Input/1M
    $10.00
    Output/1M
    $50.00
    Release
    Jun 2026

    Mythos-class model with safety restrictions for general use. State-of-the-art on nearly all benchmarks.

  • Claude Opus 4.8 Current
    Context
    1.000K
    Input/1M
    $5.00
    Output/1M
    $25.00
    Release
    May 2026

    Successor to Opus 4.7. Four times less likely to miss code bugs. Adaptive thinking automatically adjusts reasoning depth. Fast mode (research preview): 2.5x higher output speed. Mid-conversation system messages and lower prompt cache threshold (1,024 tokens). Dynamic workflows for hundreds of parallel subagents.

  • Claude Mythos 5 Current
    Context
    1.000K
    Input/1M
    $10.00
    Output/1M
    $50.00
    Release
    Jun 2026

    Unrestricted version of Fable 5, only available to cyberdefenders via Project Glasswing.

  • Claude Sonnet 4.6 Current
    Vision Thinking Tool use Computer Use Cache Batch
    Context
    1.000K
    Input/1M
    $3.00
    Output/1M
    $15.00
    Release
    Dec 2025

    The workhorse of the Claude family. Good price-to-performance for daily work, 1M context at standard pricing, works well with Claude Code and agentic workflows.

  • Claude Haiku 4.5 Current
    Vision Tool use Cache Batch
    Context
    200K
    Input/1M
    $1.00
    Output/1M
    $5.00
    Release
    Oct 2025

    Anthropic's fastest and cheapest option. For high-volume tasks like classification, extraction, customer support and simple agent loops.

  • Claude Opus 4.7 Legacy
    Vision Thinking Tool use Computer Use Cache Batch
    Context
    1.000K
    Input/1M
    $5.00
    Output/1M
    $25.00
    Release
    Apr 2026

    Anthropic flagship model. Sharpest vision (up to 3.75 megapixels) and new xhigh effort setting between high and max. Built-in cybersecurity abuse blocking. New tokenizer can produce up to 35% more tokens per input than 4.6.

Google logo

Google

6 models
  • Gemini 3.5 Flash Current
    text images audio video tool-use function-calling structured-output code-execution search-as-tool dynamic-thinking
    Context
    1.049K
    Input/1M
    $1.50
    Output/1M
    $9.00
    Release
    May 2026

    Four times faster than frontier models. Outperforms Gemini 3.1 Pro on coding and agentic benchmarks. Powers Gemini Spark under the hood. Available in GitHub Copilot.

  • Gemini 2.5 Pro Current
    Vision Audio Video Thinking Tool use Cache
    Context
    1.049K
    Input/1M
    $1.25
    Output/1M
    $10.00
    Release
    Mar 2025

    Google flagship, strong in multimodal (video and audio). Under 200K tokens costs $1.25/$10; above that, pricing doubles to $2.50/$15 for the entire prompt.

  • Gemini 2.5 Flash Current
    Vision Audio Video Tool use Cache
    Context
    1.049K
    Input/1M
    $0.30
    Output/1M
    $2.50
    Release
    Jun 2025

    The budget workhorse variant. Same multimodal features as Pro, but lower accuracy on complex tasks.

  • Gemini 3.1 Pro Preview
    Vision Thinking Multimodaal Tool use
    Context
    1.049K
    Input/1M
    $2.00
    Output/1M
    $12.00
    Release
    Apr 2026

    Preview release. Leads GPQA Diamond and HLE as of April 2026. Source: openrouter.ai.

  • Gemini 3 Pro Preview
    Vision Thinking Multimodaal Tool use
    Context
    —
    Input/1M
    —
    Output/1M
    —
    Release
    Mar 2026

    Preview release. Pricing and exact context window not yet publicly disclosed.

  • Gemini 3 Flash Preview
    Vision Multimodaal Snel
    Context
    1.049K
    Input/1M
    $0.50
    Output/1M
    $3.00
    Release
    Mar 2026

    Preview release from the Gemini 3 series. Source: openrouter.ai.

xAI logo

xAI

3 models
  • Grok 4 Current
    Thinking Tool use Real-time search
    Context
    256K
    Input/1M
    $3.00
    Output/1M
    $15.00
    Release
    Jul 2025

    Reasoning-heavy model with built-in X/Twitter search. Source: openrouter.ai.

  • Grok 4 Fast Current
    Tekst Snel
    Context
    2.000K
    Input/1M
    $0.20
    Output/1M
    $0.50
    Release
    Mar 2026

    Budget variant with 2M context window. Source: openrouter.ai.

  • Grok 3 Legacy
    Tekst Tool use
    Context
    131K
    Input/1M
    $3.00
    Output/1M
    $15.00
    Release
    Feb 2025

    Previous generation Grok, replaced by Grok 4. Source: openrouter.ai.

DeepSeek logo

DeepSeek

1 model
  • DeepSeek V3.2 Current
    Tool use Thinking Open source
    Context
    128K
    Input/1M
    $0.27
    Output/1M
    $0.42
    Release
    Dec 2025

    Open source Chinese model. Excellent price-to-performance ratio, up to 90% cheaper than Western competitors. Mixture-of-experts architecture with 671B parameters, 37B active per token.

Meta logo

Meta

1 model
  • Llama 4 Maverick Current
    Vision Tool use Open source
    Context
    1.000K
    Input/1M
    —
    Output/1M
    —
    Release
    Jan 2026

    Open-weight model. API pricing depends on the host (Together AI, Fireworks, Groq, Replicate). Check your host for current rates.

Recent AI news

AI agrees with you 50% more than humans do. Here is how to fix that.

Claude agents can now run on a schedule without a server

Anthropic pulled its SDK billing overhaul on the morning it took effect

Anthropic is at the table with Trump to get Fable 5 back

Europe is leaving $600 billion in AI productivity on the table

Your Claude automations now cost per token