Foundation Models

    AI Models Directory

    Compare foundation models by benchmarks, context window, pricing, and modality.

    GP

    GPT-5

    Proprietary

    OpenAI

    The most capable model from OpenAI with advanced reasoning, vision, and function calling.

    Parameters: ~1.8T (est.)
    Context: 128K tokens
    Benchmark: 92/100
    Released: Mar 2025
    Text
    Image
    Audio
    Visit

    $5 / 1M input, $15 / 1M output

    GE

    Gemini 2.5 Pro

    Proprietary

    Google

    Google's flagship multimodal model excelling at reasoning, code, and long-context tasks.

    Parameters: Undisclosed
    Context: 1M tokens
    Benchmark: 91/100
    Released: Jan 2025
    Text
    Image
    Audio
    Video
    Visit

    $3.50 / 1M input, $10.50 / 1M output

    WH

    Whisper V4

    Open Source

    OpenAI

    State-of-the-art speech recognition and transcription across 100+ languages.

    Parameters: 1.5B
    Context: N/A
    Benchmark: 91/100
    Released: Jan 2025
    Audio
    Visit

    $0.006 / minute

    CL

    Claude 4 Opus

    Proprietary

    Anthropic

    Anthropic's most powerful model with strong safety features and exceptional reasoning.

    Parameters: Undisclosed
    Context: 200K tokens
    Benchmark: 90/100
    Released: Jun 2025
    Text
    Image
    Visit

    $15 / 1M input, $75 / 1M output

    GK

    Grok-3

    Proprietary

    xAI

    xAI's frontier model trained with real-time data access and strong reasoning.

    Parameters: Undisclosed
    Context: 128K tokens
    Benchmark: 88/100
    Released: Feb 2025
    Text
    Image
    Visit

    $5 / 1M input, $15 / 1M output

    DE

    DALL-E 4

    Proprietary

    OpenAI

    OpenAI's latest image generation model with photorealistic output and precise prompt following.

    Parameters: Undisclosed
    Context: N/A
    Benchmark: 88/100
    Released: May 2025
    Image
    Visit

    $0.04 / image (1024x1024)

    CS

    Claude 4 Sonnet

    Proprietary

    Anthropic

    Balanced model from Anthropic with strong coding and analytical abilities at lower cost.

    Parameters: Undisclosed
    Context: 200K tokens
    Benchmark: 87/100
    Released: May 2025
    Text
    Image
    Visit

    $3 / 1M input, $15 / 1M output

    LL

    Llama 4 405B

    Open Source

    Meta

    Meta's largest open-source model competitive with frontier closed-source models.

    Parameters: 405B
    Context: 128K tokens
    Benchmark: 86/100
    Released: Apr 2025
    Text
    Image
    Visit

    Free (self-hosted)

    ML

    Mistral Large 3

    Proprietary

    Mistral

    European AI champion with strong multilingual capabilities and efficient inference.

    Parameters: 123B
    Context: 128K tokens
    Benchmark: 85/100
    Released: Feb 2025
    Text
    Image
    Visit

    $2 / 1M input, $6 / 1M output

    SO

    Sora 2

    Proprietary

    OpenAI

    OpenAI's video generation model producing cinema-quality clips from text prompts.

    Parameters: Undisclosed
    Context: N/A
    Benchmark: 85/100
    Released: Jun 2025
    Video
    Visit

    $0.10 / second of video

    SD

    Stable Diffusion 4

    Open Source

    Stability AI

    The leading open-source image generation model with extensive community ecosystem.

    Parameters: 8B
    Context: N/A
    Benchmark: 84/100
    Released: Mar 2025
    Image
    Visit

    Free (self-hosted)

    DS

    DeepSeek V3

    Open Source

    DeepSeek

    Chinese open-source model rivaling frontier models at a fraction of the compute cost.

    Parameters: 671B (37B active)
    Context: 128K tokens
    Benchmark: 84/100
    Released: Jan 2025
    Text
    Visit

    $0.27 / 1M input, $1.10 / 1M output

    QW

    Qwen 3 72B

    Open Source

    Alibaba

    Alibaba's flagship open-source model with strong multilingual and coding capabilities.

    Parameters: 72B
    Context: 128K tokens
    Benchmark: 83/100
    Released: Apr 2025
    Text
    Image
    Visit

    Free (self-hosted)

    L7

    Llama 4 70B

    Open Source

    Meta

    High-performance open-source model ideal for fine-tuning and on-premise deployments.

    Parameters: 70B
    Context: 128K tokens
    Benchmark: 82/100
    Released: Apr 2025
    Text
    Visit

    Free (self-hosted)

    MX

    Mixtral 8x22B

    Open Source

    Mistral

    Mixture-of-experts model offering strong performance with efficient inference costs.

    Parameters: 141B (22B active)
    Context: 64K tokens
    Benchmark: 79/100
    Released: Dec 2024
    Text
    Visit

    Free (self-hosted)