Pyyan / Compare / Mistral Large 4 vs GPT-6 Astra
Mistral Large 4 vs GPT-6 Astra
2 of 5Language Models · verified 10 Oct 2026
Mistral Large 4Mistralpreview
GPT-6 AstraOpenAIcurrent| Specification | Mistral Large 4 | GPT-6 Astra |
|---|---|---|
| Summary | A trillion parameters in preview, with the weights promised but not shipped. | Saturates three of the benchmarks it reports, and ships disabled for enterprise. |
| Context | 1M | 1.05M |
| Input ($/Mtok) | $1.36 | $10 |
| Output ($/Mtok) | Not published | $50 |
| Licence | Not published | Proprietary |
| Max output | Not published | 128K |
| Parameters | 1.05T total, 49B or 52B active | Not published |
| Released | 6 Oct 2026 | 3 Sep 2026 |
| Model ID | mistral-large-4 | gpt-6-astra |
| Category | Language Models | Language Models |
| Official | Mistral ↗ | — |
Highlighted rows are where these differ.
Mistral Large 4
- Preview API only: weights are promised by the end of October and the licence is unpublished
- Sources disagree on active parameters, 49B in early coverage and 52B in the documentation
- 1M token context, with a 1.6B parameter vision encoder
Best for trying a European frontier model.
GPT-6 Astra
- 97.6% on FrontierMath Tier 4, 99.9% on ARC-AGI-3 and 100% on ExploitBench
- Above 272K input tokens the whole request reprices: input and cache at 2x, output at 1.5x
- Rolls out to OpenAI's Daybreak cyber programme first, and arrives switched off for enterprise
Best for hard reasoning and computer use.