Mistral Large 3
Language Available Full comparison ↗ mistral/mistral-large-3 · by Mistral AI
· decoder-only-transformer
Pricing — 1 offering(s)
About this offering: Mistral Large 3 (current Large generation; supersedes Large 2 / 24.11). La Plateforme per-million-token pricing.
Input tokens
- $0.50 / 1M tokens (input) Current 2026-06-12 → present
Output tokens
- $1.50 / 1M tokens (output) Current 2026-06-12 → present
Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.
See how Mistral Large 3 fits into a cost-aware routing setup
See how →Capability profile
How Mistral Large 3 rates across core capability dimensions, with the task-level evidence behind each rating.
SWE-bench Verified (500 human-verified GitHub issues; patch must pass the repo's own tests; no file-location hints): resolves 41.4% — fails roughly 3 issues in 5. Independent (Vals AI leaderboard, #70 of 72; mini-SWE-agent harness). Mistral's own launch materials publish no coding benchmark for Large 3.
≈128K-token context (verify for the current generation).
Strong European-language coverage.
Coding benchmarks
View the full coding benchmark leaderboard →| Benchmark | Score | Harness | Source |
|---|---|---|---|
| swe-bench-verified ↗ | 41.4% 2025-12 | mini-swe-agent | 📊 leaderboard ↗ |
| terminal-bench-2-1 | 12.0% 2026-08 | terminus-2 | 🔬 independent ↗ |
Operator guidance
Mistral's flagship — a competitive general model, notably strong on European languages. Drop to Mistral Small 4 for cost/latency.
Use cases
- Frontier-adjacent general tasks with strong multilingual (EU) coverage
- EU-data-residency-sensitive deployments
Limitations
- Context window listed as the documented Large generation (~128K) — verify
- Capability ratings are qualitative, not a single cited benchmark run