Mistral Large 4 launch report card · BenchGecko's own tests on day one
Mistral AI's Mistral Large 4 on BenchGecko's own tests: identity, refusals, knowledge cutoff and weekly drift, with language cost pending. Price: $0.7/M input and $2.1/M output at Mistral, the cheapest of 2 offers.
Mistral Large 4 by Mistral AI appeared on OpenRouter on 2026-10-06. BenchGecko runs its own tests on every new model and adds each result here when it is in:
| Test | Result | Run date | Method |
|---|---|---|---|
| Who Are You | names Mistral AI as its maker (100% of 6 answers) | 2026-10-17 | test |
| Censorship Index | refuses or dodges 0% of 39 everyday questions | 2026-11-09 | test |
| Knowledge Horizon | knows world news up to Apr 2025 | 2026-10-17 | test |
| Model Drift | weekly check · 100% of exact tasks right · names its maker in 4 of 4 · refuses 0 of 7 | 2026-11-11 | test |
Pending: Tokenizer Tax (queued for the daily Gecko Lab run)
Not scheduled: World Map (not in the World Map roster)
Price on 2026-11-14: $0.7/M input and $2.1/M output at Mistral, the cheapest of 2 offers. All providers: https://benchgecko.ai/model/mistral-large-4-0
How it compares with every other model: Gecko Scorecard.
Published automatically when a BenchGecko test or price tracker crossed a fixed threshold. Every number and quote comes from the stored run, linked above. Spotted an error? hello@benchgecko.ai