Cloud Providers

Cloud providers require an API key and incur per-token costs. Set keys in Settings → LLM or via environment variables.

Anthropic (Claude)

Model prefix: claude-

{ "anthropic_key": "sk-ant-..." }

Model	Best for
claude-opus-4-7	Most capable — complex reasoning, long context
claude-sonnet-4-6	Balanced — quality + speed
claude-3-5-sonnet-20241022	Strong general purpose
claude-haiku-4-5-20251001	Fast and cheap — routing, classification

---

OpenAI (GPT)

Model prefixes: gpt-, o1-, o3-

{ "openai_key": "sk-..." }

Model	Best for
gpt-4o	Flagship — vision + text
gpt-4o-mini	Fast and cheap
o1	Deep reasoning
o3-mini	Fast reasoning

---

Google Gemini

Model prefixes: gemini-, gemma-

{ "gemini_key": "AIza..." }

Model	Best for
gemini-2.0-flash	Fast, large context
gemini-1.5-pro	Multimodal, 1M token context
gemma-3-27b-it	Open-weight via Gemini API

---

xAI (Grok)

Model prefix: grok-

{ "grok_key": "xai-..." }

Model	Description
grok-3	Latest Grok model
grok-2-vision	Multimodal

---

DeepSeek

Model prefix: deepseek-

{ "deepseek_key": "sk-..." }

Model	Description
deepseek-chat	General purpose (DeepSeek-V3)
deepseek-reasoner	Deep reasoning (DeepSeek-R1)

---

AWS Bedrock

Model prefix: bedrock.

AWS Bedrock provides access to multiple foundation models (Anthropic, Meta, Mistral, Amazon) through a unified AWS API.

{
  "bedrock_api_key": "absk-...",
  "aws_region": "us-east-1",
  "bedrock_inference_profile": ""
}

Models are specified with the full Bedrock model ID prefixed by bedrock.:

bedrock.anthropic.claude-3-5-sonnet-20241022-v2:0
bedrock.meta.llama3-8b-instruct-v1:0
bedrock.mistral.mistral-large-2402-v1:0

Inference profiles

Some Bedrock models require an inference profile (required for cross-region inference):

{
  "bedrock_inference_profile": "us.anthropic.claude-3-5-sonnet-20241022-v2:0"
}

Embedding model

For Bedrock mode, set an embedding model for long-term memory:

{ "embedding_model": "amazon.titan-embed-text-v2:0" }

---

OpenAI-Compatible (Cloud)

For cloud providers that expose an OpenAI-compatible API: OpenRouter, Together AI, Fireworks AI, Anyscale, etc.

Model prefix: oaic.<model_name>

{
  "openai_compatible_key": "sk-or-...",
  "openai_compatible_base_url": "https://openrouter.ai/api/v1",
  "openai_compatible_models": "mistralai/mistral-7b,meta-llama/llama-3-70b-instruct"
}

Use the model: oaic.mistralai/mistral-7b

For embedding models:

{
  "openai_compatible_embed_models": "text-embedding-3-small"
}