Mistral
Mistral is supported for chat completions and embeddings. Its small and medium models sit at the cheap end of the catalog, which makes them a common choice for classification, routing, and extraction steps where a frontier model is overkill.
Connect Mistral to TokenSense
Step 1: Go to your dashboard → Settings → Providers → add your Mistral API key.
Step 2: Point your workflow at your TokenSense endpoint and use any Mistral model name. TokenSense recognises the model and routes automatically.
Step 3: Authenticate with your TokenSense API key. Costs appear in your dashboard within seconds, attributed to the workflow that made the call.
Supported models
All pricing is per 1 million tokens (USD).
| Model | Input (per 1M) | Output (per 1M) | Use for |
|---|---|---|---|
| mistral-medium-3-5 | $1.50 | $7.50 | Chat, code, and tools |
| mistral-small-2603 | $0.15 | $0.60 | Chat, classification, and tools |
| mistral-embed | $0.10 | — | Embeddings |
Use the native model IDs above. Older friendly names are not interchangeable with current IDs. Devstral 2 is deprecated; evaluate Medium 3.5 as its replacement. Benchmark your own workflows before changing models. See Medium 3.5 and Small 4. OCR and Voxtral endpoints require separate adapters.
Using Mistral in your workflows
n8n: Select a Mistral model in the TokenSense node.
Make / Zapier: Use an HTTP module pointed at your TokenSense endpoint. See the Make & Zapier guide.
Custom code: Point your OpenAI-compatible client's base URL at your TokenSense endpoint and set a Mistral model name. See the API reference.