Skip to content
// ai · models

AI models

These are the models you can use for free in Pocket Code — you just bring your own key (BYOK). Pocket Code never sells AI credits or proxies your requests: you connect your own account and your usage is billed by the provider directly. Pick any model below across every supported provider.

Setup is two lines: paste an API key from OpenAI, Claude, Gemini or GitHub Copilot — or point Pocket Code at any OpenAI-compatible endpoint (OpenRouter, a local model, your own server) — and every model that provider offers is ready to use.

Last updated: 2026-06-15

Claude

3
  • Claude Opus 4.8thinking

    Flagship — deepest reasoning & analysis

    claude-opus-4-8Max tokens: 32,000
  • Claude Sonnet 4.6thinking

    Balanced — fast & highly capable

    claude-sonnet-4-6Max tokens: 32,000
  • Claude Haiku 4.5

    Ultrafast — great for quick tasks

    claude-haiku-4-5-20251001Max tokens: 8,192

Gemini

2
  • Gemini 3 Prothinking

    Strongest reasoning & code

    gemini-3-proMax tokens: 65,536
  • Gemini 3 Flashthinking

    Fast with thinking — balanced speed & quality

    gemini-3-flashMax tokens: 65,536

OpenAI

2
  • GPT-5

    Flagship — best for complex coding & reasoning

    gpt-5Max tokens: 32,768
  • GPT-5 Mini

    Fast & affordable — strong at code

    gpt-5-miniMax tokens: 32,768

GitHub Copilot

2
  • GPT-5

    Flagship via Copilot — complex coding & reasoning

    gpt-5Max tokens: 32,768
  • GPT-5 Mini

    Fast & affordable via Copilot

    gpt-5-miniMax tokens: 32,768