Supported models

Model ids the AI Gateway accepts, and what each one is for

Pass any of these strings as model on the SDK call:

Model	Use case
`claude-opus-4-8`	Most capable; longer reasoning, agentic workflows.
`claude-sonnet-4-6`	Workhorse; balanced cost / quality. Good default.
`claude-haiku-4-5`	Fast and cheap; tool-use loops, classification.
`nova-pro`	Amazon Nova; balanced general use with low latency.
`nova-lite`	Amazon Nova; cheaper and faster for high-volume calls.
`nova-micro`	Amazon Nova; lowest-cost text tier.
`mistral-large-3`	Mistral flagship; strong general reasoning and coding.
`pixtral-large`	Mistral vision-language; accepts image + text input.
`qwen3-coder`	Qwen coding specialist; code generation and refactoring.
`qwen3-vl`	Qwen vision-language; accepts image + text input.
`gpt-5-6-sol`	Most capable GPT; frontier reasoning, agentic coding.
`gpt-5-6-terra`	Balanced GPT for production; code, extraction, agents.
`gpt-5-6-luna`	Fast and cheap GPT; classification, summarizing, routing.
`gpt-5-5`	Previous GPT generation; general chat and reasoning.
`gpt-5-4`	Previous GPT generation; general chat and reasoning.
`titan-embed-text-v2`	Text embeddings (RAG, vector search).

All chat models are called the same way: swap the model string.

Capability by model

Capability	Models
Chat completions	Every model except `titan-embed-text-v2`
Embeddings	`titan-embed-text-v2`
Image input	`pixtral-large`, `qwen3-vl`, `gpt-5-6-*`
Web search	`gpt-*` only
Native Anthropic API	`claude-*`, under `bedrock/` ids

Next steps

Chat completions: call any chat model
Errors and limits: what the gateway doesn’t offer