DeepSeek¶

Autobot supports DeepSeek as an LLM provider via its OpenAI-compatible API. This gives access to DeepSeek-V3, DeepSeek-R1, and other DeepSeek models known for strong reasoning and coding performance at competitive pricing.

Setup¶

1. Get an API key¶

Create an API key at platform.deepseek.com.

2. Configure credentials¶

Add your API key to the .env file:

DEEPSEEK_API_KEY=...

Or use the interactive setup:

autobot setup
# Select "DeepSeek" as provider

3. Configure the provider¶

In config.yml:

agents:
  defaults:
    model: "deepseek/deepseek-chat"

providers:
  deepseek:
    api_key: "${DEEPSEEK_API_KEY}"

4. Verify¶

autobot doctor
# Should show: ✓ LLM provider configured (deepseek)

Model naming¶

Models use the deepseek/ prefix followed by the DeepSeek model ID:

# General-purpose (DeepSeek-V3)
model: "deepseek/deepseek-chat"

# Reasoning (DeepSeek-R1)
model: "deepseek/deepseek-reasoner"

The deepseek/ prefix tells autobot to route to the DeepSeek API. It is stripped before sending to the API.

Reasoning content¶

DeepSeek-R1 (the deepseek-reasoner model) returns reasoning traces in its responses. Autobot captures these in the reasoning_content field, allowing downstream features to access the model's chain-of-thought.

Custom API base¶

To use a proxy or alternative endpoint:

providers:
  deepseek:
    api_key: "${DEEPSEEK_API_KEY}"
    api_base: "https://your-proxy.example.com/v1/chat/completions"

Configuration reference¶

Field	Required	Default	Description
`api_key`	Yes	—	DeepSeek API key
`api_base`	No	`https://api.deepseek.com/v1/chat/completions`	Custom API endpoint
`extra_headers`	No	—	Additional HTTP headers for every request

How it works¶

DeepSeek uses the OpenAI-compatible Chat Completions API format:

Authorization: Bearer header for authentication
Standard message format with role and content fields
reasoning_content field returned for reasoning models (DeepSeek-R1)

Autobot detects DeepSeek models by the deepseek keyword and routes to the correct endpoint automatically. Tools, MCP servers, plugins, and all other features work the same as with other providers.

Voice transcription¶

DeepSeek does not provide a transcription API. If you need voice message support, configure an additional Groq or OpenAI provider for Whisper-based transcription.

Known limitations¶

No streaming — Responses are returned in full after the model finishes generating.
Tool choice is always auto — There is no configuration to force a specific tool or disable tool use per-request.

Troubleshooting¶

Enable debug logging to see request/response details:

LOG_LEVEL=DEBUG autobot agent -m "Hello"

Look for:

POST https://api.deepseek.com/v1/chat/completions model=... — confirms provider is active
Response 200 (N bytes) — confirms API response
HTTP 4xx/5xx: ... — API errors with details

Common issues¶

"No LLM provider configured" — Check that api_key is set and non-empty in config.yml.

"API error: authentication failed" — Invalid or expired API key. Verify at platform.deepseek.com.

"API error: rate limit exceeded" — Too many requests. DeepSeek applies per-key rate limits based on your plan.