Note: Overall leaderboard rankings may not reflect true model quality — individual benchmarks give a clearer picture. ARC-Challenge MMLU GPQA GSM8K Artificial Analysis Intelligence Index v4.1
← Back to leaderboard

Tencent: hy4-preview

5 benchmarks
CompanyTencent 腾讯控股有限公司
Country of originChina

Leaderboard position

#37
Overall rank (of 1122 models)
2702
Elo rating
5
Benchmarks
Top 3.3%
Percentile
#2
In Tencent (of 12 models)
Above: OpenAI: GPT-5.5 #36 Below: inclusionAI: ling-3.1-flash #38

Benchmark highlights

Best benchmark
GPQA Diamond 92.3
Worst benchmark
Agent Arena - Net Improvement (25.09.2026) 4.37

Nearest competitor from another company

The model from another company with the most similar benchmark scores.

appeared as nearest in 1 of 5 benchmarks

Anthropic: Claude Opus 4.8
1/5 20.0%
DeepSeek: deepseek-v4-flash-vision-exp
1/5 20.0%
reflection: beam-1.0
1/5 20.0%
Qwen: qwen3.8-flash-next
1/5 20.0%
OpenAI: GPT-5.5
1/5 20.0%
Model Votes % of benchmarks
Anthropic: Claude Opus 4.8 1 20.0%
DeepSeek: deepseek-v4-flash-vision-exp 1 20.0%
reflection: beam-1.0 1 20.0%
Qwen: qwen3.8-flash-next 1 20.0%
OpenAI: GPT-5.5 1 20.0%

Benchmarks

GPQA Diamond 92.3 SWE-bench Multilingual 82.9 SWE-bench Pro 65.7 DeepSWE 64.3 Agent Arena - Net Improvement (25.09.2026) 4.37

Predicted scores

Estimated from 25 similar models and the global Elo rating. These are projections, not measured results.

Benchmark Predicted Confidence Based on
Humanity's Last Exam (no tools) ~44.39 61% Ornith: ornith-1.5-397b, Anthropic: Claude Opus 4.8, Anthropic: claude-opus-4.5
Artificial Analysis Intelligence Index v5.0.1 ~37.42 60% Qwen: qwen3.8-27b
MATH-500 ~99.2 60% Ornith: ornith-1.5-9b
AIME 2026 ~99.2 60% Anthropic: claude-opus-4.5, Z.ai: GLM 5.2, Qwen: qwen3.6-27b
HMMT Feb 2026 ~96.9 60% Anthropic: claude-opus-4.5, Z.ai: GLM 5.2, Qwen: qwen3.6-27b
MMLU-Pro ~90.53 60% Anthropic: claude-opus-4.5, Qwen: qwen3.6-27b, Qwen: Qwen3.5 27B
IFEval ~96.75 60% Qwen: Qwen3.5 27B, Nex AGI: Nex-N2 Pro, Nex AGI: Nex-N2 Mini
IFBench ~83.97 60% Anthropic: Claude Opus 4.8, Anthropic: Claude Fable 5, Z.ai: GLM 5.2
Terminal-Bench v2.1 ~85.31 60% Ornith: ornith-1.5-397b, Anthropic: Claude Opus 4.8, Anthropic: Claude Fable 5
SWE-bench Verified ~95 60% Ornith: ornith-1.5-397b, Anthropic: claude-opus-4.5, Anthropic: Claude Fable 5
SciCode ~55.85 60% Anthropic: Claude Opus 4.8, Anthropic: Claude Fable 5, NVIDIA Corporation: nvidia-nemotron-3.5-super-vl-120b-a12b
AA-LCR ~79.29 60% Anthropic: Claude Opus 4.8, Anthropic: Claude Fable 5
Humanity's Last Exam (with tools) ~59.74 60% Ornith: ornith-1.5-397b, Anthropic: Claude Opus 4.8, Anthropic: Claude Fable 5
LiveCodeBench v6 ~90.33 60% Anthropic: claude-opus-4.5, Qwen: qwen3.6-27b, Qwen: qwen3.8-27b
MMMU-Pro ~87.69 60% Anthropic: Claude Fable 5, Qwen: qwen3.6-27b, Qwen: Qwen3.5 27B
Artifical Analysis Intelligence Index v5.0 ~37.02 60% Qwen: qwen3.8-27b
Artificial Analysis Intelligence Index v4.3.2 ~41.39 60% Qwen: qwen3.8-flash-next, Qwen: qwen3.6-27b
CowQA ~97.22 60% Z.ai: GLM 5.1
BrowseComp ~87.05 60% Ornith: ornith-1.5-397b, Anthropic: Claude Opus 4.8, NVIDIA Corporation: nvidia-nemotron-3.5-super-vl-120b-a12b
GPQA ~94.3 60% Z.ai: GLM 5.2
SuperGPQA ~73.75 60% Anthropic: claude-opus-4.5, Qwen: qwen3.6-27b, Qwen: Qwen3.5 27B
ParseBench ~77.93 60% Qwen: qwen3.8-27b
Artificial Analysis Intelligence Index v4.1.1 (Maximum Reasoning) ~55.39 60% Anthropic: Claude Opus 4.8, Anthropic: Claude Fable 5, Z.ai: GLM 5.2
Terminal-Bench v2.0 ~74.7 60% Anthropic: claude-opus-4.5, Qwen: qwen3.6-27b, Qwen: Qwen3.5 27B
ScreenSpot-Pro ~79 60% Qwen: Qwen3.5 27B