Documentation / Reference
Providers and models Every provider Seshat supports, with the models it knows, their context window and their capabilities.
This page is generated from the provider registry of the runtime. How to set keys and endpoints is in Configuration , and the code side in Providers and authentication .
Models change often Providers retire and add models all the time. The list below is what the runtime ships with. You can use any model name your provider accepts, and ollama models are whatever you have pulled locally.
Providers
Provider ID Service Authentication Vision Prompt cache anthropicAnthropic api_key Yes Yes openaiOpenAI api_key Yes Yes mistralMistral AI api_key Yes No geminiGoogle Gemini api_key Yes No deepseekDeepSeek api_key Yes No ollamaOllama none No No openrouterOpenRouter api_key Yes No codexCodex oauth No No z-aiZ.ai api_key Yes Yes minimaxMiniMax api_key Yes Yes bedrockAWS Bedrock aws Yes Yes vertexGoogle Cloud Vertex AI gcp Yes Yes foundryAzure AI Foundry api_key Yes Yes workers-aiCloudflare Workers AI api_key No No opencodeOpenCode Zen api_key Yes No kimiKimi (Moonshot AI) api_key Yes Yes
Models
Anthropic (anthropic)
Model ID Context Max output Prompt cache claude-sonnet-4-20250514200K 64K Yes claude-3-5-sonnet-20241022200K 8K Yes claude-3-5-haiku-20241022200K 8K Yes
OpenAI (openai)
Model ID Context Max output Prompt cache gpt-5.5272K 32K Yes gpt-5.4-mini272K 16K Yes gpt-4o128K 16K Yes
Mistral AI (mistral)
Model ID Context Max output Prompt cache mistral-large-latest131K 16K No mistral-small-latest131K 8K No open-mistral-7b32K 4K No
Google Gemini (gemini)
Model ID Context Max output Prompt cache gemini-2.0-pro2M 8K No gemini-2.0-flash1M 8K No gemini-1.5-flash1M 8K No
DeepSeek (deepseek)
Model ID Context Max output Prompt cache deepseek-v4-flash1M 393K No deepseek-v4-pro1M 393K No deepseek-v4-flash-vision-exp1M 393K No
OpenRouter (openrouter)
Model ID Context Max output Prompt cache anthropic/claude-3.5-sonnet200K 8K No openai/gpt-4o128K 16K No deepseek/deepseek-r164K 32K No
Codex (codex)
Model ID Context Max output Prompt cache gpt-5.6-sol200K 100K No gpt-5.5200K 100K No gpt-5.4-mini272K 16K No
Z.ai (z-ai)
Model ID Context Max output Prompt cache glm-5.1200K 128K Yes glm-4.71M 8K No glm-4.5200K 8K No
MiniMax (minimax)
Model ID Context Max output Prompt cache MiniMax-M2.7204K 128K Yes MiniMax-M2.5204K 128K Yes MiniMax-M2.1204K 8K No
AWS Bedrock (bedrock)
Model ID Context Max output Prompt cache anthropic.claude-3-5-sonnet-20241022200K 8K Yes anthropic.claude-3-5-haiku-20241022200K 8K Yes anthropic.claude-3-opus-20240229200K 4K Yes
Google Cloud Vertex AI (vertex)
Model ID Context Max output Prompt cache claude-3-5-sonnet@20241022200K 8K Yes claude-3-5-haiku@20241022200K 8K Yes claude-3-opus@20240229200K 4K Yes
Azure AI Foundry (foundry)
Model ID Context Max output Prompt cache claude-3-5-sonnet-20241022200K 8K Yes claude-3-5-haiku-20241022200K 8K Yes claude-3-opus-20240229200K 4K Yes
Cloudflare Workers AI (workers-ai)
Model ID Context Max output Prompt cache @cf/meta/llama-3.1-70b-instruct128K 4K No @cf/deepseek-ai/deepseek-r146K 4K No @cf/qwen/qwen2.5-coder-7b32K 4K No
OpenCode Zen (opencode)
Model ID Context Max output Prompt cache claude-sonnet-4200K 64K No gpt-5.3-codex200K 100K No glm-5.1200K 128K No
Kimi (Moonshot AI) (kimi)
Model ID Context Max output Prompt cache kimi-k31M 1M Yes kimi-k2.7-code262K 262K Yes kimi-k2.7-code-highspeed262K 262K Yes kimi-k2.6262K 262K Yes kimi-k2.5262K 262K Yes
Updated on 2026-10-07
