Qwen3-14B ranks first on MathIF.
Qwen Apr 2025 Open weights
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 | |
|---|---|---|---|---|
| MathIF | 50.7% | #1 of 12 | Official leaderboard | |
| ALRAGE | 79.7% | #6 of 39 | Official leaderboard | |
| Vectara Hallucination Leaderboard (2025 dataset) | 5.40% | #17 of 104 | Official leaderboard | |
| Vectara Hallucination Leaderboard (HHEM) | 2.20% | #28 of 148 | Official leaderboard | |
| Arena-Hard-Auto | 91.7% | #8 of 39 | Lab model card | |
| RULER | 94.6% | #9 of 42 | Lab model card | |
| Global PIQA | 77.1% | #13 of 44 | Official leaderboard | |
| MadinahQA | 52.8% | #13 of 39 | Official leaderboard | |
| AlignBench | 8.56 | #5 of 15 | Lab model card | |
| Multi-IF | 74.8% | #10 of 28 | Lab model card | |
| MLogiQA | 71.1% | #3 of 8 | Lab model card | |
| ArbMMLU-HT | 47.8% | #16 of 39 | Official leaderboard | |
| Open Arabic LLM Leaderboard v2 | 53.8% | #19 of 39 | Official leaderboard | |
| AlGhafa | 57.1% | #20 of 39 | Official leaderboard | |
| Fiction.LiveBench | 62.5% | #30 of 57 | Fiction.live leaderboard, via Epoch AI | |
| ArabicMMLU | 44.6% | #22 of 39 | Official leaderboard | |
| Encyclo-K | 30.8% | #13 of 23 | Official leaderboard | |
| MT-AIME2024 | 73.3% | #4 of 7 | Lab model card | |
| IFEval | 85.4% | #78 of 135 | Lab model card | |
| Creative Writing v3 | 80.3 | #7 of 12 | Lab model card | |
| SEA-HELM | 64.8 | #42 of 71 | Official leaderboard | |
| WritingBench | 78.0 | #9 of 14 | Lab model card | |
| INCLUDE | 71.7% | #20 of 31 | Lab model card | |
| MMLU-Redux | 88.6% | #30 of 45 | Lab model card | |
| HABIT-Bench | 31.4% | #4 of 6 | Official leaderboard | |
| C-Eval | 86.2% | #31 of 46 | Lab model card | |
| MMMLU | 77.9% | #50 of 71 | Lab model card | |
| KINA | 26% | #27 of 37 | Official leaderboard | |
| AraGen | 29.7% | #32 of 42 | Official leaderboard | |
| DiscoX | 39.4 | #6 of 7 | Official leaderboard | |
| Longform Creative Writing (EQ-Bench) | 35.9 | #103 of 119 | Official leaderboard | |
| Alyah | 26% | #9 of 10 | Official leaderboard | |
| Meeseeks Utility Rate | 58% | #10 of 10 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| AutoLogi | 89.2% | #2 of 8 | Lab model card | |
| MATH-500 | 96.8% | #23 of 70 | Lab model card | |
| ZebraLogic | 88.5% | #10 of 23 | Lab model card | |
| AIME 2024 | 79.3% | #44 of 88 | Lab model card | |
| OTIS Mock AIME 2024-2025 | 66.4% | #118 of 212 | Epoch AI run | |
| GPQA Diamond | 63.8% | #133 of 229 | Epoch AI run | |
| DTBench | 64% | #128 of 180 | Official leaderboard | |
| PolyMath | 45.8% | #18 of 24 | Lab model card | |
| AIME 2025 | 70.4% | #112 of 143 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LiveCodeBench v5 | 63.5% | #21 of 47 | Lab model card | |
| ArtifactsBench (Gemini 2.5 Pro 0506 judge) | 39.8% | #9 of 18 | Official leaderboard | |
| Multi-LCB | 42.4% | #7 of 11 | Official leaderboard | |
| Codeforces rating | 1,766 | #33 of 47 | Lab model card | |
| AutoCodeBench | 37.6% | #24 of 34 | Official leaderboard | |
| GitTaskBench | 5.56% | #9 of 10 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| C3-Bench | 9.38% | #8 of 30 | Official leaderboard | |
| Berkeley Function Calling Leaderboard (BFCL) | 70.4% | #16 of 41 | Lab model card | |
| AppWorld Test-Normal | 86.9% | #5 of 12 | Official leaderboard | |
| BFCL V4 | 41% | #28 of 65 | Official leaderboard | |
| AppWorld Test-Challenge | 67.6% | #6 of 12 | Official leaderboard | |
| PlanBench-XL | 0% | #6 of 9 | Official leaderboard | |
| SnakeBench | 20 | #137 of 188 | Official leaderboard | |
| AdaPlanBench | 17.3% | #9 of 10 | Official leaderboard | |
| LisanBench | 58.3 | #120 of 133 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| BizFinBench | 67% | #5 of 13 | Official leaderboard | |
| BRIDGE | 46.8 | #20 of 39 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MolecularIQ | 13.1% | #6 of 17 | Official leaderboard | |
| AthenaBench | 42.7% | #3 of 6 | Official leaderboard | |
| AraTrust | 57.5% | #28 of 39 | Official leaderboard | |
| nocot-bench (NCRI) | 55.0 | #156 of 185 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LiveBench | 71.3% | #11 of 27 | Lab model card | |
| ACCoRD | 48.4 | #116 of 152 | Official leaderboard | |
| Conceptual Reasoning Index | 28.6 | #118 of 151 | Official leaderboard | |
| LMCA | 18.2% | #116 of 147 | Official leaderboard |
The Self-Reported Index fits one score to the 18 capability results Qwen 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 132.2 to 135.2. 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 |
|---|---|---|---|
| Berkeley Function Calling Leaderboard (BFCL) | 70.4% BFCL v3, thinking mode (Table 15). Non-thinking 61.5 (Table 16) | 68% Leaderboard, native function calling | +2.4 pts |
| GPQA Diamond | 64% GPQA-Diamond, thinking mode (Table 15). Non-thinking 54.8 (Table 16) | 63.8% Epoch AI run | +0.2 pts |
| PolyMath | 45.8% PolyMath, thinking mode (Table 15). Non-thinking 22.0 (Table 16) | 45.8% Leaderboard, thinking | Same |
| RULER | 94.6% RULER Avg. (4K to 128K, YaRN x4), non-thinking mode, 128K 85.1. Thinking mode Avg. 90.1 with 8192-token thinking budget (Table 23) | 94.6% Leaderboard | Same |
Qwen's models in color, against every other model on the Epoch Capabilities Index, by release date.
Source: Epoch Capabilities Index, Epoch AI (CC BY 4.0)
Every result in the longer documents the lab published with it. 25 of them aren't in the launch post.