At a Glance
| Vendor | Mistral AI |
| Family | Mistral Large |
| Launched | Mistral launched Mistral Large 4 in public preview on October 6, 2026 and said weights would follow by the end of the month. Mistral's docs now describe it as open-weight. We could not confirm a live weight download on October 31, so check the model page directly. |
| Context window | 1M tokens, up from 256K on Large 3. It is a mixture-of-experts model with 1.05 trillion total parameters, 52 billion active, and a 1.6 billion parameter vision encoder, taking text and image input. |
| Pricing | Mistral's docs list $0.70 per million input tokens, $0.07 cached, and $2.10 output as of October 31. Coverage from October 23 gave $1.36 / $4.18, so the listed price has changed or differs by tier. Use Mistral's page as the reference. |
| Access channels | Mistral API and Mistral's own products. Weights are expected on Hugging Face, subject to the check above. |
Notable Benchmarks
Vendor-reported results via launch coverage: 61.7 percent on DeepSWE v1.1, 28.3 percent on Terminal-Bench 4, 93 percent on Cybench, and 42 percent on Dense 200 against 41 percent for GPT-6 Astra. Coverage also reports a top-five place on the Artificial Analysis Cyber Index. Terminal-Bench 4 puts it well behind Kimi K3.1 Pro and Claude Opus 5.5.
Strengths
A European frontier-scale model with vision, 1M context, and low API pricing. It suits organisations that need a non-US, non-Chinese provider.
Limitations
The license is the Mistral Ecosystem Open Weight License v26.10, not the Apache 2.0 used for Large 3, so review its terms. Self-hosting is demanding: one community estimate is about 500 GB at 4-bit for the weights alone. It trails the leaders on terminal coding tests.
Brand-Visibility Implications
Large 4 matters most for European buyers, where sovereignty rules push enterprises toward Mistral and its Le Chat assistant. Brands selling into France, Germany, and regulated EU sectors should add it to tracked models. French and German sources are likely to matter more here than in US-centric assistants, which is our inference and worth testing with your own prompts. See Mistral revenue and ARR.
How Presenc AI Tracks This Model
Presenc AI monitors brand visibility on Mistral AI's Mistral Large as part of continuous multi-platform AI visibility tracking. We sample Mistral Large 4 across representative prompt sets daily, compare against competitor performance on the same prompts, and flag material mention-rate changes so brand teams can respond quickly when AI representation shifts.