Other · model
ERNIE 5.5 Flash Preview benchmark scores
As of 2026-11-18, ERNIE 5.5 Flash Preview (Other) has measured scores on 14 benchmarks and estimated scores on 78 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 91.5% | estimated ± 2.7 pp, high confidence |
| AA-HLE | 45.2% | estimated ± 4.6 pp, high confidence |
| AA-SciCode | 54.4% | estimated ± 3.1 pp, high confidence |
| Artificial Analysis Intelligence Index | 37.3% | estimated ± 5.7 pp, medium confidence |
| CritPt | 20.3% | estimated ± 5.5 pp, medium confidence |
| GDPval-AA | 57.4% | estimated ± 4.6 pp, medium confidence |
| GPQA Diamond | 94.1% | estimated ± 2.0 pp, high confidence |
| HLE | 54.2% | estimated ± 6.5 pp, medium confidence |
| LiveCodeBench v6 | 96.1% | estimated ± 1.9 pp, medium confidence |
| AA-Omniscience Accuracy | 45.5% | estimated ± 7.8 pp, medium confidence |
| SciCode | 60.1% | estimated ± 7.6 pp, low confidence |
| BrowseComp | 87.4% | measured |
| HLE w/ tools | 59.8% | estimated ± 5.6 pp, medium confidence |
| VITA-Bench | 46.7% | estimated ± 6.7 pp, low confidence |
| τ²-bench results | 93.5% | estimated ± 11.8 pp, low confidence |
| AA Coding Index | 76.5% | estimated ± 2.5 pp, high confidence |
| APEX-Agents | 64.2% | measured |
| APEX-Agents-AA | 36.8% | estimated ± 4.4 pp, high confidence |
| DeepSearchQA | 90.2% | estimated ± 7.0 pp, low confidence |
| SWE-bench Verified | 82.2% | estimated ± 2.3 pp, medium confidence |
| AutomationBench | 61.9% | measured |
| BFCL v4 | 77.2% | estimated ± 4.7 pp, low confidence |
| CyberGym | 89.3% | estimated ± 4.6 pp, medium confidence |
| JobBench | 58.3% | estimated ± 8.0 pp, medium confidence |
| SkillsBench | 59.9% | estimated ± 4.1 pp, medium confidence |
| SWE-bench Pro | 62.6% | estimated ± 3.5 pp, medium confidence |
| WideResearch | 81.2% | estimated ± 3.5 pp, high confidence |
| AA Agentic Index | 51.8% | estimated ± 3.3 pp, medium confidence |
| MMLU-Pro | 85.8% | estimated ± 2.5 pp, low confidence |
| GPQA | 93.5% | estimated ± 3.3 pp, high confidence |
| Vals GPQA Diamond | 91.6% | estimated ± 2.9 pp, high confidence |
| Vals LiveCodeBench | 77.4% | estimated ± 6.1 pp, low confidence |
| Vals MMLU-Pro | 89.2% | estimated ± 1.8 pp, high confidence |
| Chess Puzzles | 56.6% | estimated ± 8.0 pp, low confidence |
| Gert Labs | 73.0% | estimated ± 7.1 pp, low confidence |
| AA-AnalystAgent | 45.0% | estimated ± 8.0 pp, low confidence |
| AA Harvey LAB v1.0 | 92.3% | estimated ± 9.8 pp, low confidence |
| AA Tau3 Banking | 52.0% | estimated ± 3.3 pp, low confidence |
| ApprenticeBench | 25.5% | estimated ± 11.0 pp, low confidence |
| ARC-AGI-1 | 95.2% | estimated ± 3.9 pp, high confidence |
| ARC-AGI-2 | 77.7% | estimated ± 13.0 pp, low confidence |
| CursorBench 3.2 | 30.2% | estimated ± 3.6 pp, low confidence |
| FrontierCode 1.1 Main | 50.4% | estimated ± 3.5 pp, low confidence |
| MazeBench | 3.9% | estimated ± 4.0 pp, high confidence |
| Mystery Game Puzzles | 61.0% | estimated ± 12.6 pp, low confidence |
| NYT Connections | 88.2% | estimated ± 6.4 pp, low confidence |
| OSWorld-Verified | 81.6% | estimated ± 5.7 pp, medium confidence |
| Terminal-Bench 4.1 | 45.4% | estimated ± 5.0 pp, medium confidence |
| Vals SWE-bench | 98.4% | estimated ± 5.2 pp, low confidence |
| CWE-Bench-AA | 68.6% | estimated ± 6.0 pp, low confidence |
| AA Cyber Index | 46.3% | estimated ± 9.4 pp, low confidence |
| DeepsecBench-AA | 20.2% | estimated ± 5.2 pp, low confidence |
| HealthBench Professional (length-adjusted, Artificial Analysis) | 58.1% | estimated ± 4.4 pp, medium confidence |
| HealthBench Professional (raw, Artificial Analysis) | 65.9% | estimated ± 2.6 pp, medium confidence |
| AA ITBench | 49.0% | estimated ± 4.1 pp, high confidence |
| Bug Hunt Bench | 32.0% | estimated ± 7.2 pp, low confidence |
| CursorBench 4.0 | 47.4% | estimated ± 4.7 pp, high confidence |
| DeepSWE | 71.2% | measured |
| DRACO | 84.9% | measured |
| HLE w/o tools | 46.4% | measured |
| OSWorld 2.0 | 19.6% | estimated ± 5.5 pp, low confidence |
| PostTrainBench v1.1 | 39.2% | estimated ± 4.5 pp, high confidence |
| ProgramBench | 29.5% | measured |
| Terminal-Bench 4.0 | 46.5% | measured |
| Terminal-Bench-Science 0.1 | 21.0% | measured |
| Toolathlon-Verified | 75.0% | measured |
| VulcanBench Frontier v4 | 88.5% | estimated ± 1.2 pp, medium confidence |
| AA Terminal-Bench 4.0 | 43.5% | estimated ± 6.3 pp, medium confidence |
| ARC-AGI-3 | 7.5% | estimated ± 7.2 pp, medium confidence |
| CWE-bench | 62.9% | estimated ± 11.0 pp, low confidence |
| MKA | 37.2% | estimated ± 10.6 pp, low confidence |
| Hermes Bench | 71.1% | estimated ± 6.4 pp, low confidence |
| Hermes Index | 42.2% | estimated ± 10.2 pp, low confidence |
| ExploitGym | 21.0% | estimated ± 9.2 pp, low confidence |
| Claw-Eval | 73.1% | estimated ± 6.3 pp, low confidence |
| DeepPlanning | 52.2% | estimated ± 13.1 pp, low confidence |
| MCP Atlas | 84.6% | measured |
| NL2Repo | 64.8% | measured |
| QwenClawBench | 69.8% | estimated ± 2.6 pp, medium confidence |
| Toolathlon | 57.6% | estimated ± 4.2 pp, high confidence |
| MedXpertQA (Text) | 63.7% | estimated ± 8.9 pp, low confidence |
| ResearchClawBench | 19.8% | estimated ± 1.8 pp, medium confidence |
| HealthBench Professional | 55.1% | estimated ± 4.2 pp, medium confidence |
| ScarfBench | 67.2% | estimated ± 3.8 pp, medium confidence |
| BlenderWorld (text-only) | 79.0% | measured |
| Cookie-Bench | 88.2% | measured |
| Agents' Last Exam | 36.8% | estimated ± 11.6 pp, low confidence |
| OpenHarmony Bench v1.0 | 59.8% | estimated ± 2.7 pp, high confidence |
| SWE-Marathon | 45.0% | estimated ± 12.9 pp, low confidence |
| LifeVal | 77.9% | estimated ± 1.6 pp, medium confidence |
| PinchBench | 90.6% | estimated ± 3.6 pp, low confidence |
| SWE-Atlas Codebase QnA | 60.9% | estimated ± 4.8 pp, high confidence |