Models

Put the model name from the tables below into the model field of your request body. You can switch models on the same endpoint at any time. The tables list the models the gateway currently offers; actual availability depends on your account in the console.

Query Models in Real Time

Call GET /v1/models (with your key) at any time to get the list of models your key can currently access.

Anthropic Claude

Supports both the OpenAI-compatible protocol (/v1/chat/completions) and the Claude-native protocol (/v1/messages).

ModelNotes
claude-opus-4-8Opus flagship, strongest reasoning
claude-opus-4-7Opus
claude-opus-4-6Opus
claude-opus-4-5-20251101Opus (dated version)
claude-opus-4-1-20250805Opus (dated version)
claude-sonnet-4-6Sonnet, balanced speed and quality
claude-sonnet-4-5-20250929Sonnet (dated version)
claude-haiku-4-5Haiku, lightweight and fast
claude-haiku-4-5-20251001Haiku (dated version)

OpenAI GPT

ModelNotes
gpt-5.5General-purpose chat
gpt-5.4 / gpt-5.4-high / gpt-5.4-miniGeneral-purpose chat, different tiers
gpt-5.2 / -low / -medium / -high / -xhighDifferent reasoning-effort tiers
gpt-5.3-codex / -low / -medium / -high / -xhighCode-focused model
codex-auto-reviewAutomated code review

Google Gemini

OpenAI-compatible calls use the table below; for the native protocol see Gemini Native.

ModelNotes
gemini-3.1-pro-previewPro preview
gemini-3-pro-previewPro preview
gemini-3-flash-previewFlash preview
gemini-3-pro-image-previewPro image understanding preview
gemini-3.1-flash-image-previewFlash image preview
gemini-2.5-pro2.5 Pro
gemini-2.5-flash / gemini-2.5-flash-thinking2.5 Flash

DeepSeek

ModelNotes
deepseek-v4-proHigh performance
deepseek-v4-flashLightweight and fast

Image Generation

Called through the image endpoint, billed per image.

ModelNotes
nano-banana-proImage generation (high quality)
nano-banana-2Image generation
nano-bananaImage generation
gpt-image-2 / gpt-image-2-vipImage generation

Video Generation

ModelNotes
sora-2-proVideo generation (high quality)
sora-2Video generation