Mistral AI & Codestral
Integrate Mistral's sophisticated model suite, including the specialized Codestral engine, into Sypha.
Mistral AI & Codestral
Sypha supports the full spectrum of Mistral AI models, ranging from the high-capacity Mistral Large to the specialized Codestral engine, purpose-built for high-fidelity code generation.
Official Site: mistral.ai
Obtaining Your API Keys
- Console Access: Register or authenticate at La Plateforme (Mistral Console).
- Credential Generation:
- Standard API Keys for general-purpose models.
- Codestral-Specific Keys for specialised coding instances.
Supported Model Registry
Sypha integrates with several key Mistral architectures:
| Engine Identifier | Function Calling | Vision Capabilities | Typical Use Case |
|---|---|---|---|
codestral-latest | Yes | No | Specialised code synthesis and refactoring. |
mistral-large-latest | Yes | No | High-complexity reasoning and logic. |
pixtral-large-latest | Yes | Yes | Multi-modal tasks involving visual data. |
mistral-small-latest | Yes | No | Efficient, low-latency interactions. |
[!TIP] While Sypha defaults to a temperature of 0.0 for precision, we suggest experimenting with manual temperature adjustments for creative tasks.
Configuring Sypha
- Activate Settings: Access the gear icon within the Sypha interface.
- Select Registry: Choose "Mistral" from the API Provider menu.
- Insert Credentials: Paste your Mistral API key into the relevant field.
- Define Model: Select your preferred engine from the list.
Utilising the Codestral Engine
Codestral is engineered specifically for software development. Depending on your key type, you may need to adjust the endpoint:
- Codestral Dedicated Endpoint: Defaulting to
codestral.mistral.ai. - Standard Platform Access: If using a La Plateforme key for Codestral, update the Codestral Base URL to
https://api.mistral.ai.
Note: Service availability and specific model parameters are subject to Mistral AI’s current Model Overview.