Auto
Cheap Flash-class for short chat; stronger when the run gets long or tool-heavy.
Models
What this is: the full Gateway catalog your desk can reach. Who it is for: anyone who refuses a single-lab lock-in. Day one: Auto routes short chat to Fast and heavy runs to Strong. Approvals: model choice never bypasses send approval. Cost: Gateway tokens on your meter.
Fast
Short answers, light tools. Default: Gemini Flash-class.
Strong
Planning, long context, many tool calls. Claude-class default.
Pinned
Any Gateway slug you name. Fail-over stays inside the same class.
Gateway families — live ids load when the catalog is reachable.
OpenAI
Fastopenai/gpt-4.1-miniopenai/gpt-4.1
Anthropic
Stronganthropic/claude-sonnet-4.5anthropic/claude-sonnet-4
google/gemini-2.5-flashgoogle/gemini-2.5-pro
xAI
Pinnedxai/grok-3-minixai/grok-3
Meta
Pinnedmeta/llama-4-maverickmeta/llama-3.3-70b
Mistral
Pinnedmistral/mistral-smallmistral/mistral-large
DeepSeek
Pinneddeepseek/deepseek-chatdeepseek/deepseek-reasoner
Amazon
Pinnedamazon/nova-liteamazon/nova-pro
Cohere
Pinnedcohere/command-rcohere/command-r-plus
Full catalog
Loading Gateway models…
Provider cards above show family samples while the list loads.