AI Models
The model catalogue

Every great open model. One key.

The foundational open-weight families the industry builds on, ready the moment you sign up — and a proprietary architecture that makes each of them sharper.

Start free Try the playground
Models available
Dozens
Leading families
5+
API for all of them
1
Free tokens / month
2M
Families

The families the world builds on

Open-weight models from the most respected research teams, curated, tuned and kept current for you.

  • Llama family

    Versatile general-purpose models with excellent instruction following, from fast compact sizes to frontier-scale reasoning.

  • Qwen family

    Strong multilingual and coding performance with long context — ideal for global products and developer tools.

  • Mistral family

    Efficient, low-latency models that punch above their weight for chat, extraction and real-time apps.

  • Gemma family

    Lightweight, well-mannered models that shine on focused tasks and high-volume workloads.

  • DeepSeek family

    Reasoning-first models built for math, code and multi-step problem solving.

  • And more, continuously

    New families and versions arrive in the catalogue as they earn their place. No migrations, no new contracts.

Intelligent routing

Stop guessing which model to use

Set model to “auto” and our routing layer reads each request and sends it to the model best suited to the task — balancing quality, speed and cost on every call.

  • Task-aware selection, request by request
  • Automatic fallback keeps you online
  • Pin a specific model whenever you want full control
request.ts
// 1. Create a key in your dashboard → API keys
// 2. Copy your base URL from the same screen
const res = await fetch(`${BASE_URL}/api/v1/chat/completions`, {
  method: 'POST',
  headers: {
    Authorization: `Bearer ${process.env.AI_MODELS_KEY}`,
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
    model: 'auto',            // or any model id from /api/v1/models
    messages: [{ role: 'user', content: 'Hello, world' }],
    stream: true,
  }),
});
Beyond raw models

Proprietary architecture, built in

  • Routing

    The right model for every request, chosen in milliseconds.

  • Memory

    Context that persists across sessions, scoped to user, team or workspace.

  • Skills

    Reusable capabilities — search, read, write, call — any model can use.

  • Agents

    Goal-driven workers that plan, act and report back.

Pick by job

A family for every workload

A quick guide. Or let auto-routing decide for you.

CapabilityChatCodeReasoningMultilingualHigh volume
Llama family
Qwen family
Mistral family
Gemma family
DeepSeek family
From zero to shipped

Live in four steps

  1. 01

    Sign up free

    No card. 2M tokens a month from day one.

  2. 02

    Try in the playground

    Compare families side by side on your own prompts.

  3. 03

    Grab a key

    One key unlocks every model in the catalogue.

  4. 04

    Ship

    Stream responses into your product in minutes.

FAQ

Model questions, answered

Which models can I use on the Free plan?

Every open model in the catalogue. Plans differ by included tokens, throughput and team features — not by which models you can reach.

Do I have to choose a model?

No. Use “auto” and routing will pick for you. You can pin a specific model any time.

How quickly do new models arrive?

We add families and versions as they prove themselves on quality and reliability, and announce them in the changelog.

Will my data be used to train models?

No. Your prompts and outputs are never used to train shared models.

Meet every model in five minutes

Sign up, open the playground and put the catalogue to work on your own problems.

Start free See pricing