Mistral Small 2503
The mistral-small model is a cost-efficient reasoning model designed for low-latency workloads. It leverages the Mixtral-8X7B-v0.1 architecture, which incorporates a sparse mixture of experts with 12B active parameters. With improved reasoning capabilities, the model demonstrates a broad range of functionalities, including code generation and reasoning. It also supports multiple languages such as English, French, German, Italian, and Spanish, making it a versatile option for various linguistic tasks.
- Input / output price
- $0.351 / $0.555 per 1M
- Context
- 128K tokens
- Released
- Mar 17, 2025
Paste into Claude Code, Cursor, or Codex
Help me set up the Respan gateway: fetch and follow https:/
Providers
One provider serves this model, priced per 1M tokens after gateway discounts.
| Provider | Input | Output |
|---|---|---|
Concentrate concentrate/mistral-small-3.1 | $0.351/M | $0.555/M |
Pricing
USD per 1M tokens on Concentrate, after gateway discounts.
- Input
- $0.351 per 1M
- Output
- $0.555 per 1M
Cost by volume
Total cost from 0 to 1T tokens.
Throughput and latency
30-day averages from requests through the Respan gateway. Under each figure, every model on the gateway by its average, this one in blue.
Output speed
Higher is better13tokens/sacross providers
Response latency
Lower is better1.8sacross providers
Benchmarks
8 published scores from 1 source, each linked to where it was published.
Published scores
| Benchmark | Score | Source | Published |
|---|---|---|---|
| HumanEvalhuggingface.co | 88.4% | huggingface.co | Mar 17, 2025 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| GPQA Diamondhuggingface.co | 46.0% | huggingface.co | Mar 17, 2025 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| MATHhuggingface.co | 69.3% | huggingface.co | Mar 17, 2025 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| MMLUhuggingface.co | 80.6% | huggingface.co | Mar 17, 2025 |
| MMLU-Prohuggingface.co | 66.8% | huggingface.co | Mar 17, 2025 |
| SimpleQAhuggingface.co | 10.4% | huggingface.co | Mar 17, 2025 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| MMMUhuggingface.co | 64% | huggingface.co | Mar 17, 2025 |
| MMMU-Prohuggingface.co | 49.3% | huggingface.co | Mar 17, 2025 |
Apps
Use Mistral Small 2503 in the tools you already run. Each guide points the app at the Respan gateway.
Setup guides for more apps are in the integration docs.
Mistral Small 2503 FAQ
Short answers, built from the data on this page.
Mistral Small 2503 is a model from Mistral, released on Mar 17, 2025. The mistral-small model is a cost-efficient reasoning model designed for low-latency workloads. It leverages the Mixtral-8X7B-v0.1 architecture, which incorporates a sparse mixture of experts with 12B active parameters. With improved reasoning capabilities, the model demonstrates a broad range of functionalities, including code generation and reasoning. It also supports multiple languages such as English, French, German, Italian, and Spanish, making it a versatile option for various linguistic tasks.
On the Respan gateway, Mistral Small 2503 costs $0.351 per 1M input tokens and $0.555 per 1M output tokens on Concentrate, after gateway discounts.
Mistral Small 2503 is served by Concentrate.
Mistral Small 2503 accepts up to 128K tokens of context.
Point any OpenAI-compatible SDK at https://gateway.respan.ai/v1 with your Respan API key and set the model to "mistral-small-2503". To set it up with a coding agent, copy the agent setup prompt at the top of this page.
Mistral Small 2503 has 8 published benchmark scores, including HumanEval 88.4%, GPQA Diamond 46.0%, and MATH 69.3%. Each score links to where it was published; the data is collected by Respan.
Certified where it counts.
Built to meet the security and privacy standards that enterprise and healthcare teams require.
ISO 27001
The internationally recognized standard for information security management.
SOC 2
Secure, compliant management of your data across all of our systems.
GDPR
Operated under GDPR, the world's strictest standard for data privacy.
HIPAA
HIPAA compliant, with a BAA available for healthcare teams.