Hunyuan-A13B-Instruct places in the top three on 2 benchmarks.
Tencent Jun 2025 Open weights Mid-size
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| BIG-Bench Hard | 89.1% | #8 of 73 | Lab model card | |
| DROP | 91.1% | #7 of 41 | Lab model card | |
| MMLU-Pro | 67.3% | #94 of 212 | Official leaderboard | |
| LongBench v2 | 55% | #21 of 35 | Lab model card | |
| IFEval | 84.7% | #85 of 135 | Lab model card | |
| FRAMES | 81.1% | #6 of 9 | Lab model card | |
| SysBench | 76.1% | #4 of 6 | Lab model card | |
| RULER | 76.7% | #35 of 42 | Lab model card | |
| ComplexNLU (Tencent) | 61.2% | #1 of 1 | Lab model card | |
| InsCtrl (Tencent) | 71.9% | #1 of 1 | Lab model card | |
| LengthCtrl (Tencent) | 55.4% | #1 of 1 | Lab model card | |
| PenguinScrolls | 87.7% | #1 of 1 | Lab model card | |
| Word-Task (Tencent) | 62.9% | #1 of 1 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MATH | 94.3% | #6 of 99 | Lab model card | |
| AIME 2024 | 87.3% | #27 of 88 | Lab model card | |
| OlympiadBench | 82.7% | #1 of 3 | Lab model card | |
| GPQA Diamond | 71.2% | #174 of 302 | Lab model card | |
| ZebraLogic | 84.7% | #14 of 23 | Lab model card | |
| AIME 2025 | 76.8% | #100 of 143 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| FullStackBench | 67.8% | #2 of 8 | Lab model card | |
| ArtifactsBench (Gemini 2.5 Pro 0506 judge) | 43% | #6 of 18 | Official leaderboard | |
| LiveCodeBench | 63.9% | #20 of 54 | Lab model card | |
| ArtifactsBench | 43% | #4 of 5 | Lab model card | |
| McEval | 59.6% | #1 of 1 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Berkeley Function Calling Leaderboard (BFCL) | 78.3% | #3 of 41 | Lab model card | |
| ComplexFuncBench | 61.2% | #1 of 3 | Lab model card | |
| tau-bench | 54.7% | #20 of 23 | Lab model card | |
| SnakeBench | 15 | #169 of 188 | Official leaderboard | |
| C3-Bench | 63.5% | #1 of 1 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| nocot-bench (NCRI) | 48.7 | #169 of 185 | Official leaderboard | |
| LiveSecBench | 16.9 | #34 of 34 | Official leaderboard |
The Self-Reported Index fits one score to the 15 capability results Tencent reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 134.2 to 138.5. How we compute it.
Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.
| Benchmark | Lab says | Board says | Gap |
|---|---|---|---|
| ArtifactsBench | 43% ArtifactsBench | 41% Leaderboard | +2.0 pts |
| ComplexNLU (Tencent) | 61.2% ComplexNLU | 61.2% Leaderboard | Same |
The benchmarks the lab chose to headline, with the scores it gave.
Every result in the longer documents the lab published with it. 5 of them aren't in the launch post.