Other · model
LLaDA2.2-mini benchmark scores
As of 2026-11-22, LLaDA2.2-mini (Other) has measured scores on 9 benchmarks and estimated scores on 30 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 55.3% | estimated ± 4.4 pp, medium confidence |
| AA-HLE | 0.0% | estimated ± 3.9 pp, medium confidence |
| AA-SciCode | 4.5% | estimated ± 4.7 pp, medium confidence |
| AIME26 | 35.1% | measured |
| CritPt | 0.0% | estimated ± 4.6 pp, medium confidence |
| GDPval-AA | 0.0% | estimated ± 5.5 pp, low confidence |
| GPQA Diamond | 44.4% | measured |
| HLE | 6.3% | estimated ± 6.8 pp, low confidence |
| IFBench | 24.9% | measured |
| AA-LCR | 33.6% | estimated ± 12.5 pp, low confidence |
| LiveCodeBench v6 | 28.1% | measured |
| AA-Omniscience Accuracy | 2.8% | estimated ± 8.4 pp, low confidence |
| SciCode | 7.7% | estimated ± 4.9 pp, medium confidence |
| BrowseComp | 55.5% | estimated ± 13.0 pp, low confidence |
| IFEval | 77.7% | estimated ± 3.6 pp, medium confidence |
| LongBench v2 | 35.0% | measured |
| VITA-Bench | 35.0% | estimated ± 10.7 pp, low confidence |
| τ²-bench results | 57.5% | measured |
| AA Coding Index | 2.7% | estimated ± 5.7 pp, low confidence |
| APEX-Agents-AA | 1.6% | estimated ± 11.4 pp, low confidence |
| SWE-bench Verified | 57.5% | estimated ± 9.7 pp, low confidence |
| BFCL v4 | 47.7% | measured |
| JobBench | 23.7% | estimated ± 9.3 pp, low confidence |
| AA Agentic Index | 11.5% | estimated ± 12.7 pp, low confidence |
| MMLU-Pro | 55.5% | estimated ± 5.4 pp, medium confidence |
| AA-IFBench | 56.7% | estimated ± 7.4 pp, low confidence |
| GPQA | 44.1% | estimated ± 1.2 pp, high confidence |
| MATH-500 | 89.5% | estimated ± 3.1 pp, low confidence |
| AIME 2025 | 39.2% | estimated ± 1.8 pp, low confidence |
| OSWorld-Verified | 67.5% | estimated ± 3.8 pp, medium confidence |
| Claw-Eval | 57.2% | measured |
| HMMT Feb 2026 | 16.8% | estimated ± 4.0 pp, medium confidence |
| MCP Atlas | 46.0% | estimated ± 13.5 pp, low confidence |
| MCP-Tasks | 68.1% | estimated ± 6.4 pp, low confidence |
| MMLU-Redux | 80.7% | estimated ± 2.8 pp, high confidence |
| SuperGPQA | 23.7% | estimated ± 13.9 pp, low confidence |
| MMLU-Pro (Arcee) | 63.9% | estimated ± 3.4 pp, low confidence |
| ResearchClawBench | 15.7% | estimated ± 1.3 pp, high confidence |
| PinchBench | 62.3% | measured |