Mistral Small 2603
At a glance
- Price per 1M tokens (input)
- $0.15
- Price per 1M tokens (output)
- $0.60
- Context
- 262,144 tokens
- Free access
- yes (see below)
Refreshed daily; data verified September 29, 2026. Published August 12, 2026, last price change August 30, 2026.
Mistral Small 2603 is a compact language model designed for teams that need reliable image understanding, tool calling (the ability to invoke external functions or APIs), and step-by-step reasoning without the overhead of a larger system. It suits production workflows such as document analysis, automated customer support, and multi-step data processing where accuracy and speed matter. The vendor emphasizes efficiency and control, offering a model that balances capability with deployability across standard infrastructure. Its approach focuses on practical utility rather than benchmark chasing, making it a solid choice for businesses integrating AI into existing software stacks.
Specifications & pricing
| Input (per 1M tokens) | $0.15 |
|---|---|
| Output (per 1M tokens) | $0.60 |
| Cache read (per 1M tokens) | $0.015 |
| Context window | 262,144 tokens |
| Max output | 262,144 tokens |
| Capabilities | images, tool calling, reasoning |
LiteLLM community dataset (MIT), verified September 29, 2026. Official Mistral AI pricing.
What Mistral Small 2603 would cost on your workload — run it through the cost calculator →
Price history
| Date | Input / 1M | Output / 1M |
|---|---|---|
| August 12, 2026 (first record) | $0.15 | $0.60 |
| August 30, 2026 | $0.15 | $0.60 |
Where to try Mistral Small 2603 for free
- Mistral AI offers a free chat — Le Chat (free plan). A vendor's free chat may run a different model from the same family — the exact model is not guaranteed.
Frequently asked questions
What is this model best used for in a business context?+
It is ideal for tasks that combine visual and textual information, like extracting data from invoices or screenshots, and for automating workflows that require calling external tools, such as database lookups or sending messages. Its step-by-step reasoning also helps with complex problem-solving, like troubleshooting or planning, where you need transparent logic.
How do we get started with integrating this model?+
You can access it through the Mistral AI platform or via API, where you can test it with sample prompts. For production, you can deploy it on your own infrastructure using the provided model weights and standard inference frameworks. Start with a small pilot project, like a document classification task, to evaluate performance and latency.
How does this model differ from other models in the Mistral line?+
Compared to larger siblings, this model is optimized for lower latency and easier deployment while retaining essential capabilities like image understanding and tool calling. It may have less depth in very complex reasoning or broader knowledge than premium models, but it offers a cost-effective balance for routine business tasks. It also differs from text-only models by natively handling images, which expands its use cases.
What are the limitations we should be aware of?+
The model may struggle with highly nuanced visual details or very long, multi-step reasoning chains. It is not designed for tasks requiring deep domain expertise or creative generation. Also, tool calling requires careful setup of your function schemas, and the model may occasionally misinterpret instructions, so you should implement validation and fallback logic in your workflow.
Compare with others
Org chart: how to move your company onto AI
A practical map: which company roles and processes AI agents can take over, where to start, and in what order to roll it out.