Mistral Provider

Mistral AI is a European AI company producing high-performance models with strong coding abilities. Their models offer a good balance of quality and speed.

Get an API Key

Go to console.mistral.ai
Create an account or sign in
Navigate to API Keys
Create a new key and copy it

Configure VibeCody

Option 1: Environment variable (recommended)

export MISTRAL_API_KEY="..."
vibecli --provider mistral

Option 2: Config file (~/.vibecli/config.toml)

[mistral]
enabled = true
api_key = "..."
model = "mistral-large-latest"

Model Selection

Model	Strengths	Best for
`mistral-large-latest`	Strongest reasoning	Complex coding tasks
`mistral-medium-latest`	Good balance	General-purpose coding
`mistral-small-latest`	Fast, affordable	Quick completions
`codestral-latest`	Code-specialized	Code generation, completion

Default: mistral-large-latest

Override from the CLI:

vibecli --provider mistral --model codestral-latest

Codestral

Codestral is Mistral’s code-specialized model, trained specifically for code generation, completion, and explanation. Use it for pure coding tasks where you want maximum code quality.

Pricing

Check mistral.ai/pricing for current rates. Mistral is competitively priced compared to other frontier providers.

Best For

European data sovereignty – Mistral is based in the EU
Code-specialized models – Codestral for pure coding tasks
Competitive pricing – strong quality at lower cost than GPT-4o or Claude

Verify Connection

vibecli --provider mistral -c "Write a Rust struct with serde serialization"

Troubleshooting

Invalid API key

Error: 401 Unauthorized

Check your key at console.mistral.ai
Confirm the env var is set: echo $MISTRAL_API_KEY

Model not found

Use mistral-large-latest (not just mistral-large)
Check docs.mistral.ai for current model names

Provider: Mistral