GLM-5.3
GLM-5.3 is Z.AI's open-weight flagship for complex coding, long-horizon agentic work, and cybersecurity tasks. It uses the same base model as GLM-5.2, with its capability gains coming from post-training. The Hugging Face checkpoint is a text-only 753B-parameter MoE (about 40B active) with 78 layers, 256 routed experts plus one shared expert, eight routed experts selected per token, DeepSeek Sparse Attention with IndexShare, and a 1,048,576-token context window. It supports up to 128K output tokens, function calling, structured output, context caching, and always-on thinking with low, high, and max reasoning-effort levels. The downloadable FP8 and BF16 weights are available under the custom GLM-5.3 License, which permits use, modification, redistribution, sale, deployment, and fine-tuning but requires Model-as-a-Service businesses above US$10B in annual revenue to pass a Z.AI security review before commercial use.
Identité
| Organisation | Zhipu AI |
|---|---|
| Pays | — |
| Famille | — |
| Type | llm |
| Date de sortie | 2026-08-14 |
| Licence | glm-5.3 license |
| Licence commerciale | oui |
| Licence open source | non |
| Poids ouverts | oui |
| Auto-hébergement (DGX Spark) | non — 753 Md > 107 Md |
| Modalités | text |
| Fenêtre de contexte | 1 048 576 |
| KV cache / token (Q8) | 1.92 Mo/token |
| Paramètres totaux | 753 B |
| Paramètres actifs (session) | 40 B |
| Source des paramètres actifs | web : https://github.com/zai-org/GLM-5 |
| Curation des paramètres actifs | 2026-11-04 |
| Prix par session | 2.00 $ (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) | 132 M |
| Ratio Terminal-Bench / Data Players | 109.8× |
| Coût / tâche (Terminal-Bench) | 40.91 $ |
| Tokens d'entraînement | — |
| Coupure de connaissance | — |
| Créé le | 2026-08-14T06:49:20.874750Z |
| Mis à jour le | 2026-11-04T06:05:14.148755Z |
| Source | LLM Stats |
| URL | https://llm-stats.com/models/glm-5.3 |
IA Act — preuves fournisseur
| Critère | Note / 5 | Poids |
|---|---|---|
| Documentation technique | 2,5 | 5 % |
| Résumé des données d'entraînement | 0,0 | 35 % |
| Conformité droit d'auteur | 0,0 | 35 % |
| Code of Practice GPAI | 0,0 | 20 % |
| Risque systémique | 0,0 | 5 % |
Curation : 2026-10-23.
IA ActPrix par fournisseur
Statut du modèle (offres + auto-hébergement) : souverain — via AKI.IO (cortecs, eurouter), Arkane (cortecs), Entrim (cortecs, eurouter), GreenPT (eurouter), Inceptron (cortecs, eurouter), Mistral AI (cortecs), Nebul (eurouter), Scaleway (cortecs, eurouter, scaleway), Tensorix (cortecs, eurouter).
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 |
|---|---|---|---|---|---|---|---|---|---|
| AKI.IO | cortecs, eurouter | souverain | 1.12 $ | 3.93 $ | nvfp4 | 512 000 / 1 048 576 | — | — | — |
| Arkane | cortecs | souverain | 1.41 $ | 4.43 $ | — | 200 000 | — | — | — |
| Entrim | cortecs, eurouter | souverain | 0.52 $ | 2.06 $ | nvfp4 | 1 048 576 | — | — | — |
| GreenPT | eurouter | souverain | — | — | — | 1 048 576 | — | — | — |
| Inceptron | cortecs, eurouter | souverain | 0.88 $ | 3.33 $ | fp4 | 1 048 576 | — | — | — |
| Mistral AI | cortecs | souverain | 1.55 $ | 4.88 $ | — | 1 000 000 | — | — | — |
| Nebul | eurouter | souverain | — | — | — | 1 048 576 | — | — | — |
| Scaleway | cortecs, eurouter, scaleway | souverain | 2.02 $ | 6.17 $ | — | 256 000 / 1 048 576 | — | — | — |
| Tensorix | cortecs, eurouter | souverain | 1.76 $ | 4.53 $ | fp8 | 1 048 576 | — | — | — |
| Sference | cortecs | partiel | 1.21 $ | 4.23 $ | nvfp4 | 1 048 576 | — | — | — |
| DeepInfra | llm-stats | non | 1.20 $ | 4.00 $ | — | — | — | — | active |
| FriendliAI | llm-stats | non | 1.40 $ | 4.40 $ | — | — | — | — | active |
| Melious | eurouter | non | — | — | — | 1 048 576 | — | — | — |
| Novita | llm-stats | non | 1.40 $ | 4.40 $ | — | — | — | — | active |
| ZAI | llm-stats | non | 1.40 $ | 4.40 $ | — | — | — | — | active |
Souveraineté qualifiée par offre (modèle × fournisseur) ; survolez le badge pour les critères.
SouverainetéScores par catégorie (top_scores)
| Catégorie | Score |
|---|---|
| legal | 1 769.00 |
| math | 0.60 |
| reasoning | 197.00 |
| safety | 0.50 |
| finance | 1 769.00 |
| general | 104.50 |
| agents | 111.00 |
| code | 0.50 |
| systems | 0.40 |
| tool_calling | 0.50 |
| vision | 0.60 |
Scores par benchmark
| Benchmark | Catégorie | Score | Max | Normalisé (API) | Vérifié | Auto-déclaré | Daté du |
|---|---|---|---|---|---|---|---|
| agents-last-exam | agents | 0.28 | 1.00 | 0.28 | non | oui | 2026-11-04T06:05:14.190792Z |
| automationbench | agents | 0.48 | 1.00 | 0.48 | non | oui | 2026-11-04T06:05:14.187963Z |
| cybergym | code | 0.84 | 1.00 | 0.84 | non | oui | 2026-11-04T06:05:14.176303Z |
| deepswe-1.1 | code | 0.67 | 1.00 | 0.67 | non | oui | 2026-11-04T06:05:14.159000Z |
| exploitbench | code | 0.54 | 1.00 | 0.54 | non | oui | 2026-11-04T06:05:14.182212Z |
| exploitgym | agents | 0.15 | 1.00 | 0.15 | non | oui | 2026-11-04T06:05:14.179139Z |
| frontierswe | code | 0.78 | 1.00 | 0.78 | non | oui | 2026-11-04T06:05:14.167390Z |
| gdpval-aa | general | 1 769.00 | 3 000.00 | 0.59 | non | oui | 2026-11-04T06:05:14.196373Z |
| humanity's-last-exam | reasoning | 0.62 | 1.00 | 0.62 | non | oui | 2026-11-04T06:05:14.193508Z |
| nl2repo | code | 0.58 | 1.00 | 0.58 | non | oui | 2026-11-04T06:05:14.161636Z |
| posttrainbench | agents | 0.40 | 1.00 | 0.40 | non | oui | 2026-11-04T06:05:14.173409Z |
| program-bench | agents | 0.19 | 1.00 | 0.19 | non | oui | 2026-11-04T06:05:14.164616Z |
| swe-marathon | code | 0.42 | 1.00 | 0.42 | non | oui | 2026-11-04T06:05:14.170302Z |
| terminal-bench-2.1 | reasoning | 0.88 | 1.00 | 0.88 | non | oui | 2026-11-04T06:05:14.152757Z |
| terminal-bench-3.0 | reasoning | 0.28 | 1.00 | 0.28 | non | oui | 2026-11-04T06:05:14.156037Z |
| terminal-bench-4.0 | reasoning | 0.42 | 1.00 | 0.42 | non | non | 2026-11-04T06:05:14.199409Z |
| toolathlon | agents | 0.73 | 1.00 | 0.73 | non | oui | 2026-11-04T06:05:14.185147Z |
Classements
| Catégorie | Rang | Score | Note conservatrice | Benchmarks évalués | Méthode | Daté du | Provenance |
|---|---|---|---|---|---|---|---|
| agents | 21 | 111.00 | 36.93 | 16 | trueskill | 2026-11-04T08:08:47.839150Z | rankings |
| code | 17 | 0.50 | 39.87 | 12 | trueskill | 2026-11-04T08:08:39.742311Z | rankings |
| math | 9 | 0.60 | 41.96 | 1 | trueskill | 2026-11-04T08:08:42.442591Z | rankings |
| reasoning | 15 | 197.00 | 49.98 | 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