Tencent: hy3
14 benchmarks
Leaderboard position
#107
Overall rank (of 1121 models)
2224
Elo rating
14
Benchmarks
Top 9.5%
Percentile
In Tencent (of 12 models)
Above: Anthropic: claude-opus-4.5 #106
Below: H company: holo2-235b-a22b #108
Benchmark highlights
Best benchmark
BrowseComp
84.2
Worst benchmark
MathArena Apex
38.7
Nearest competitor from another company
The model from another company with the most similar benchmark scores.
appeared as nearest in 2 of 14 benchmarks
2/14
14.3%
2/14
14.3%
| Model | Votes | % of benchmarks |
|---|---|---|
| Qwen: qwen3.7-plus | 2 | 14.3% |
| Thinking Machines Lab: inkling | 2 | 14.3% |
| Ornith: ornith-1.5-35b-a3b | 1 | 7.1% |
| orcarouter: orcacyber-nano-1.0 | 1 | 7.1% |
| Anthropic: Claude Opus 4.8 | 1 | 7.1% |
| OpenAI: GPT-5.5 | 1 | 7.1% |
| LiquidAI: lfm-3-1.7b | 1 | 7.1% |
| nanbeige: nanbeige4.3-33b | 1 | 7.1% |
| Z.ai: GLM 5.1 | 1 | 7.1% |
| internscience: sage-a1.5 | 1 | 7.1% |
Benchmarks
BrowseComp
84.2
MCP-Atlas
79.1
SWE-bench Multilingual
75.8
FrontierScience-Olympiad
74.8
AA-LCR
73.4
Terminal-Bench v2.1
71.7
Claw-Eval
68.5
SWE-bench Pro
57.9
SkillsBench
55.3
Humanity's Last Exam (with tools)
53.2
NL2Repo
45.6
Artificial Analysis Intelligence Index v4.1.1 (Maximum Reasoning)
42
Artificial Analysis Intelligence Index v4.1 (Maximum Reasoning)
41
MathArena Apex
38.7
Predicted scores
Estimated from 25 similar models and the global Elo rating. These are projections, not measured results.