Mistral Large 4
Mistral Large 4 (aka ML4 / Le Chonk) is Mistral AI's flagship open-weight, general-purpose multimodal Mixture-of-Experts model, unifying instruction following, reasoning, coding, and agentic workflows in a single set of weights. Mistral's Hugging Face card describes it as a 1T-parameter model with 49B active parameters per token (52B including embeddings and output layers; docs list 1.05T total with a 1.6B vision encoder), 128 routed experts (4 active) plus 1 shared expert, a 1M-token context window, text+image input and text output, and three reasoning efforts (low, medium, high default). It entered public preview on 2026-10-06; on 2026-10-29 Mistral released the weights of an improved, further-RL-trained version (same API id mistral-large-4) under the Mistral Ecosystem Open-Weight License (modified MIT; Model-as-a-Service businesses need a separate commercial license), priced at $0.70/$0.07/$2.10 per million input/cached input/output tokens.
Identité
| Organisation | Mistral AI |
|---|---|
| Pays | — |
| Famille | — |
| Type | llm |
| Date de sortie | 2026-10-06 |
| Licence | modified mit license |
| Licence commerciale | non |
| Licence open source | non |
| Poids ouverts | oui |
| Auto-hébergement (DGX Spark) | non — 1.05e+03 Md > 107 Md |
| Modalités | image|text |
| Fenêtre de contexte | 1 000 000 |
| KV cache / token (Q8) | 589.8 ko/token |
| Paramètres totaux | 1.05 T |
| Paramètres actifs (session) | 52 B |
| Source des paramètres actifs | web : https://docs.mistral.ai/models/mistral-large-4 |
| Curation des paramètres actifs | 2026-10-19 |
| Prix par session | 1.12 $ (1 000 000 tokens entrée + 200 000 sortie, prix le moins cher par fournisseur) |
| Tokens / tâche (Data Players) | 1.2 M |
| Tokens / tâche (Terminal-Bench) | 83.3 M (est.) |
| Ratio Terminal-Bench / Data Players | 69.4× |
| Coût / tâche (Terminal-Bench) | 9.41 $ (est.) |
| Tokens d'entraînement | — |
| Coupure de connaissance | — |
| Créé le | 2026-10-07T16:54:41.733217Z |
| Mis à jour le | 2026-11-04T06:04:56.723299Z |
| Source | LLM Stats |
| URL | https://llm-stats.com/models/mistral-large-4 |
IA Act — preuves fournisseur
| Critère | Note / 5 | Poids |
|---|---|---|
| Documentation technique | 3,8 | 5 % |
| Résumé des données d'entraînement | 5,0 | 35 % |
| Conformité droit d'auteur | 5,0 | 35 % |
| Code of Practice GPAI | 2,5 | 20 % |
| Risque systémique | — | 5 % |
Curation : 2026-10-21.
IA ActPrix par fournisseur
Statut du modèle (offres + auto-hébergement) : souverain — via Mistral AI (eurouter, llm_stats).
Toutes sources confondues : catalogue llm-stats, routeurs EUrouter/Cortecs et catalogues directs (OVHcloud, Scaleway, Infomaniak). Les prix sont convertis en USD pour la comparaison ; le prix publié et le taux BCE utilisé sont dans l'infobulle. Les prix inconnus restent indiqués « — ».
| Fournisseur | Source | Souveraineté | Entrée (/M tok) | Sortie (/M tok) | Quantification | Contexte | Latence (s) | Débit (tok/s) | Statut |
|---|---|---|---|---|---|---|---|---|---|
| Mistral AI | eurouter | souverain | — | — | — | 1 048 576 | — | — | — |
| Mistral AI | llm-stats | souverain | 0.70 $ | 2.10 $ | — | — | — | — | active |
Souveraineté qualifiée par offre (modèle × fournisseur) ; survolez le badge pour les critères.
SouverainetéScores par catégorie (top_scores)
| Catégorie | Score |
|---|---|
| knowledge | 0.20 |
| legal | 0.20 |
| math | 0.90 |
| multimodal | 0.40 |
| physics | 0.90 |
| productivity | 1 393.00 |
| reasoning | 155.30 |
| safety | 0.90 |
| search | 0.80 |
| finance | 0.50 |
| general | 107.70 |
| agents | 139.90 |
| biology | 0.90 |
| chemistry | 0.90 |
| code | 0.70 |
| tool_calling | 0.50 |
| vision | 0.40 |
Scores par benchmark
| Benchmark | Catégorie | Score | Max | Normalisé (API) | Vérifié | Auto-déclaré | Daté du |
|---|---|---|---|---|---|---|---|
| aa-briefcase | agents | 1 393.00 | 3 000.00 | 0.46 | non | oui | 2026-11-04T06:04:56.783355Z |
| automationbench | agents | 0.60 | 1.00 | 0.60 | non | oui | 2026-11-04T06:04:56.773193Z |
| b3-ai-security-benchmark | safety | 0.97 | 1.00 | 0.97 | non | oui | 2026-11-04T06:04:56.849741Z |
| browsecomp | reasoning | 0.84 | 1.00 | 0.84 | non | oui | 2026-11-04T06:04:56.861317Z |
| chartqapro | vision | 0.63 | 1.00 | 0.63 | non | oui | 2026-11-04T06:04:56.797443Z |
| coding-agent-index | code | 0.48 | 1.00 | 0.48 | non | oui | 2026-11-04T06:04:56.753450Z |
| cybench | safety | 0.93 | 1.00 | 0.93 | non | oui | 2026-11-04T06:04:56.829368Z |
| cybergym | code | 0.82 | 1.00 | 0.82 | non | oui | 2026-11-04T06:04:56.833927Z |
| deepswe-1.1 | code | 0.62 | 1.00 | 0.62 | non | oui | 2026-11-04T06:04:56.729699Z |
| dense200 | vision | 0.42 | 1.00 | 0.42 | non | oui | 2026-11-04T06:04:56.792488Z |
| finance-agent-v2 | agents | 0.55 | 1.00 | 0.55 | non | oui | 2026-11-04T06:04:56.818937Z |
| finch | finance | 0.73 | 1.00 | 0.73 | non | oui | 2026-11-04T06:04:56.843280Z |
| frontiercode-2.0 | reasoning | 0.20 | 1.00 | 0.20 | non | oui | 2026-11-04T06:04:56.865541Z |
| gdp-pdf | general | 0.19 | 1.00 | 0.19 | non | oui | 2026-11-04T06:04:56.809047Z |
| harveys-legal-agent-benchmark | legal | 0.16 | 1.00 | 0.16 | non | oui | 2026-11-04T06:04:56.823003Z |
| korabench | safety | 1.69 | 2.00 | 0.85 | non | oui | 2026-11-04T06:04:56.855811Z |
| scicode | reasoning | 0.92 | 1.00 | 0.92 | non | oui | 2026-11-04T06:04:56.812730Z |
| surge-ai-coding-quality | code | 3.74 | 5.00 | 0.75 | non | oui | 2026-11-04T06:04:56.763238Z |
| swe-atlas-codebase-qna | code | 0.59 | 1.00 | 0.59 | non | oui | 2026-11-04T06:04:56.733145Z |
| terminal-bench-4.0 | reasoning | 0.41 | 1.00 | 0.41 | non | oui | 2026-11-04T06:04:56.745331Z |
Classements
| Catégorie | Rang | Score | Note conservatrice | Benchmarks évalués | Méthode | Daté du | Provenance |
|---|---|---|---|---|---|---|---|
| agents | 30 | 139.90 | 34.93 | 10 | trueskill | 2026-11-04T08:08:47.839150Z | rankings |
| code | 32 | 0.70 | 37.10 | 6 | trueskill | 2026-11-04T08:08:39.742311Z | rankings |
| knowledge | 21 | 0.20 | 11.58 | 1 | trueskill | 2026-11-04T08:08:45.139626Z | rankings |
| math | 45 | 0.90 | 35.55 | 1 | trueskill | 2026-11-04T08:08:42.442591Z | rankings |
| reasoning | 42 | 155.30 | 45.12 | 9 | trueskill | 2026-11-04T08:08:41.093494Z | rankings |
Données
Dates de curation (récupération depuis l'API) : models 2026-11-04 · benchmarks 2026-11-04 · scores 2026-11-04 · updates 2026-11-04 · account 2026-11-04 · model_history 2026-11-04 · rankings 2026-11-04