DeepSeek

DeepSeek exposes an OpenAI-compatible Chat Completions endpoint and uses bearer authentication.

Add the key

export DEEPSEEK_API_KEY="your-deepseek-api-key"

Configure the provider

[model_providers.deepseek]
base_url = "https://api.deepseek.com"
api_backend = "chat_completions"
env_key = "DEEPSEEK_API_KEY"
context_window = 200000

[model.deepseek-v4-flash]
model = "deepseek-v4-flash"
model_provider = "deepseek"
name = "DeepSeek V4 Flash"

[model.deepseek-v4-pro]
model = "deepseek-v4-pro"
model_provider = "deepseek"
name = "DeepSeek V4 Pro"

Model catalogs change. Use only IDs returned for your DeepSeek account and set context_window no higher than the chosen model's documented limit.

Verify

grok \
  -p "Reply with exactly: MODEL_OK" \
  -m deepseek-v4-pro \
  --max-turns 1 \
  --no-subagents \
  --disable-web-search \
  --no-memory \
  --output-format json

If your account exposes different model IDs, update both model and the alias used with -m.

Official reference:

• DeepSeek API documentation