Self-hosting and data control
Choose gpt-oss-20b
gpt-oss-20b has open weights, so you can run it on your own servers; Magistral Small 1.2 is only available through its provider.
gpt-oss-20b vs Magistral Small 1.2: price, context and capabilities
Choose gpt-oss-20b for self-hosting and data control. Choose Magistral Small 1.2 for the latest release.
In short
- Magistral Small 1.2 has been retired; gpt-oss-20b is the one you can still call.
- A price comparison is not possible: one of them has no comparable published price.
- Magistral Small 1.2 is the newer model (September 2025, against August 2025).
- Only gpt-oss-20b offers reasoning, tool calling, self-hostable.
Which should you choose?
Each answer below comes straight from the published prices, specifications and independent benchmark results.
Latest release
Choose Magistral Small 1.2
Magistral Small 1.2 is the newer release (September 2025, against August 2025 for gpt-oss-20b).
Specifications side by side
| gpt-oss-20b | Magistral Small 1.2 | |
|---|---|---|
| Provider | OpenAI | Mistral AI |
| Category | Language models | Language models |
| Status | Generally available | Retired |
| Headline price | No data | No data |
| Input | No data | No data |
| Output | No data | No data |
| Cached input | No data | No data |
| Context window | 131k | No data |
| Max output | 131k | No data |
| Released | August 2025 | September 2025 |
| Knowledge cutoff | June 2024 | No data |
| Licence | Open weights | Proprietary |
| Capabilities | Reasoning, Tool calling, Self-hostable | No data |
| API model id | gpt-oss-20b | magistral-small-2509 |
Benchmarks side by side
Leader in bold.
| Benchmark | gpt-oss-20b | Magistral Small 1.2 |
|---|---|---|
| Intelligence Index | 9 | No data |
| Terminal-Bench Hard | 10.6% | No data |
| SciCode | 38.9% | No data |
| GPQA Diamond | 68.8% | No data |
| Humanity's Last Exam | 11% | No data |
| IFBench (instruction following) | 65.1% | No data |
| τ²-Bench (agentic tool use) | 60.2% | No data |
| Long-context reasoning (AA-LCR) | 34.7% | No data |
Independent evaluations by Artificial Analysis, run the same way for every model, at the reasoning setting shown. Percentages unless marked as an index.
Frequently asked questions
Should I choose gpt-oss-20b or Magistral Small 1.2?
Choose gpt-oss-20b for self-hosting and data control. Choose Magistral Small 1.2 for the latest release.
Which is cheaper, gpt-oss-20b or Magistral Small 1.2?
A price comparison is not possible: one of them has no comparable published price.
What is the difference between gpt-oss-20b and Magistral Small 1.2?
gpt-oss-20b is made by OpenAI and Magistral Small 1.2 by Mistral AI. Magistral Small 1.2 has been retired; gpt-oss-20b is the one you can still call. A price comparison is not possible: one of them has no comparable published price. Magistral Small 1.2 is the newer model (September 2025, against August 2025). Only gpt-oss-20b offers reasoning, tool calling, self-hostable.
Which is newer, gpt-oss-20b or Magistral Small 1.2?
Magistral Small 1.2 was released on September 2025 and gpt-oss-20b on August 2025.