Hy3
Hy3 is a 295B-parameter Mixture-of-Experts (MoE) model with 21B active parameters and a 3.8B MTP layer, developed by the Tencent Hy Team. Following the Hy3 Preview launch in late April, the team scaled up post-training with higher-quality data and RL, gathering feedback from 50+ products. Hy3 outperforms similar-size models and rivals flagship open-source models with 2-5x the parameters, with strong gains in reasoning, agentic, and long-context tasks. It uses 80 layers (plus 1 MTP layer), 64 GQA attention heads (8 KV heads, head dim 128), a 4096 hidden size, 192 experts with top-8 activated, a 256K context window, and BF16 precision. Hy3 is a hybrid-thinking model supporting configurable reasoning effort (no_think, low, high), and emphasizes production-grade tool-call and output-format stability, reduced hallucination, and reliable multi-turn intent tracking.
Identité
| Organisation | Tencent |
|---|---|
| Pays | — |
| Famille | — |
| Type | llm |
| Date de sortie | 2026-07-06 |
| Licence | apache 2.0 |
| Licence commerciale | oui |
| Licence open source | oui |
| Poids ouverts | oui |
| Auto-hébergement (DGX Spark) | non — 295 Md > 107 Md |
| Modalités | text |
| Fenêtre de contexte | 262 144 |
| KV cache / token (Q8) | 163.8 ko/token |
| Paramètres totaux | 295 B |
| Paramètres actifs (session) | 21 B |
| Source des paramètres actifs | web : https://huggingface.co/tencent/Hy3 |
| Curation des paramètres actifs | 2026-11-04 |
| Prix par session | 0.26 $ (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) | 2.15 $ (est.) |
| Tokens d'entraînement | — |
| Coupure de connaissance | — |
| Créé le | 2026-07-15T20:16:15.645501Z |
| Mis à jour le | 2026-11-04T06:05:12.594283Z |
| Source | LLM Stats |
| URL | https://llm-stats.com/models/hy3 |
Prix par fournisseur
Statut du modèle (offres + auto-hébergement) : non — hors DGX Spark (> 107 Md).
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 |
|---|---|---|---|---|---|---|---|---|---|
| DeepInfra | llm-stats | non | 0.14 $ | 0.58 $ | — | — | — | — | active |
Souveraineté qualifiée par offre (modèle × fournisseur) ; survolez le badge pour les critères.
SouverainetéScores par catégorie (top_scores)
| Catégorie | Score |
|---|---|
| knowledge | 0.50 |
| long_context | 0.70 |
| math | 4.20 |
| physics | 0.80 |
| reasoning | 1.90 |
| science | 0.60 |
| search | 0.80 |
| frontend_development | 0.80 |
| general | 1.60 |
| agents | 0.60 |
| biology | 0.90 |
| chemistry | 0.70 |
| code | 0.60 |
| coding | 0.50 |
| tool_calling | 0.70 |
Scores par benchmark
| Benchmark | Catégorie | Score | Max | Normalisé (API) | Vérifié | Auto-déclaré | Daté du |
|---|---|---|---|---|---|---|---|
| aa-lcr | reasoning | 0.73 | 1.00 | 0.73 | non | oui | 2026-11-04T06:05:12.687733Z |
| apex-agents | reasoning | 0.26 | 1.00 | 0.26 | non | oui | 2026-11-04T06:05:12.630921Z |
| arxivmath | math | 0.52 | 1.00 | 0.52 | non | oui | 2026-11-04T06:05:12.663926Z |
| browsecomp | reasoning | 0.84 | 1.00 | 0.84 | non | oui | 2026-11-04T06:05:12.615711Z |
| cl-bench | code | 0.24 | 1.00 | 0.24 | non | oui | 2026-11-04T06:05:12.681259Z |
| cl-bench-(life) | code | 0.17 | 1.00 | 0.17 | non | oui | 2026-11-04T06:05:12.684697Z |
| claw-eval | code | 0.69 | 1.00 | 0.69 | non | oui | 2026-11-04T06:05:12.634291Z |
| cmt-benchmark | physics | 0.38 | 1.00 | 0.38 | non | oui | 2026-11-04T06:05:12.672369Z |
| deepsearchqa | search | 0.91 | 1.00 | 0.91 | non | oui | 2026-11-04T06:05:12.621701Z |
| deepswe | code | 0.28 | 1.00 | 0.28 | non | oui | 2026-11-04T06:05:12.612610Z |
| frontierscience-olympiad | reasoning | 0.75 | 1.00 | 0.75 | non | oui | 2026-11-04T06:05:12.655502Z |
| frontierscience-research | reasoning | 0.21 | 1.00 | 0.21 | non | oui | 2026-11-04T06:05:12.652563Z |
| gpqa | reasoning | 0.90 | 1.00 | 0.90 | non | oui | 2026-11-04T06:05:12.649543Z |
| horizonmath | math | 0.07 | 1.00 | 0.07 | non | oui | 2026-11-04T06:05:12.666697Z |
| humanity's-last-exam-(no-tools,-text-only) | reasoning | 0.47 | 1.00 | 0.47 | non | oui | 2026-11-04T06:05:12.646502Z |
| humanity's-last-exam-(with-tools,-text-only) | reasoning | 0.53 | 1.00 | 0.53 | non | oui | 2026-11-04T06:05:12.643557Z |
| imo-answerbench | math | 0.90 | 1.00 | 0.90 | non | oui | 2026-11-04T06:05:12.675530Z |
| matharena-apex | math | 0.39 | 1.00 | 0.39 | non | oui | 2026-11-04T06:05:12.661065Z |
| mcp-atlas | reasoning | 0.79 | 1.00 | 0.79 | non | oui | 2026-11-04T06:05:12.624794Z |
| nl2repo | code | 0.46 | 1.00 | 0.46 | non | oui | 2026-11-04T06:05:12.609683Z |
| phybench | physics | 0.77 | 1.00 | 0.77 | non | oui | 2026-11-04T06:05:12.669562Z |
| skillsbench | code | 0.55 | 1.00 | 0.55 | non | oui | 2026-11-04T06:05:12.640636Z |
| superchem | chemistry | 0.55 | 1.00 | 0.55 | non | oui | 2026-11-04T06:05:12.678326Z |
| swe-bench-multilingual | reasoning | 0.76 | 1.00 | 0.76 | non | oui | 2026-11-04T06:05:12.597629Z |
| swe-bench-pro | reasoning | 0.58 | 1.00 | 0.58 | non | oui | 2026-11-04T06:05:12.603721Z |
| swe-bench-verified | reasoning | 0.78 | 1.00 | 0.78 | non | oui | 2026-11-04T06:05:12.600669Z |
| terminal-bench-2.1 | reasoning | 0.72 | 1.00 | 0.72 | non | oui | 2026-11-04T06:05:12.606571Z |
| toolathlon | agents | 0.48 | 1.00 | 0.48 | non | oui | 2026-11-04T06:05:12.627899Z |
| usamo-2026 | math | 30.24 | 42.00 | 0.72 | non | oui | 2026-11-04T06:05:12.658285Z |
| widesearch | search | 0.76 | 1.00 | 0.76 | non | oui | 2026-11-04T06:05:12.618809Z |
| wildclawbench | agents | 0.54 | 1.00 | 0.54 | non | oui | 2026-11-04T06:05:12.637486Z |
Classements
| Catégorie | Rang | Score | Note conservatrice | Benchmarks évalués | Méthode | Daté du | Provenance |
|---|---|---|---|---|---|---|---|
| knowledge | 29 | 0.50 | 4.45 | 1 | trueskill | 2026-11-04T08:08:45.139626Z | rankings |
| math | 49 | 4.20 | 34.89 | 8 | trueskill | 2026-11-04T08:08:42.442591Z | rankings |
| agents | 55 | — | 26.36 | 14 | trueskill | 2026-10-19 | model_history |
| code | 44 | — | 32.27 | 10 | trueskill | 2026-10-19 | model_history |
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