Up to four models
Qwen3.8-27B vs Gemini 3 Flash
On the index (GPT-4 = 100), Qwen3.8-27B scores 155 and Gemini 3 Flash 154. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
155139–171
WeightsOpen
ReleasedAug 2026
Price in / out$0.060 / $2.2
Context1.02M
Core tests taken3 of 9
154139–170
WeightsClosed
ReleasedDec 2025
Price in / out$0.50 / $3.0
Context1.05M
Core tests taken5 of 9
Qwen3.8-27B and Gemini 3 Flash are too close to call (1 point apart, within the uncertainty), ahead on 2 of 2 shared core tests, and costs 1.9× less.
Test by testthird-party results
Every result as a table
| Test | Qwen3.8-27B | Gemini 3 Flash |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Vals AI) | 88.9% | 87.9% |
| MMLU Pro | 84.3% | 88.6% |
| MMMU Pro | 83.9% | 87.6% |
| GeoBench | 38.0% | 88.0% |
| SimpleBench | 60.2% | 61.1% |
| Coding | ||
| Code Migration | 14.2% | 6.4% |
| LiveCodeBench | 84.0% | 85.6% |
| ProgramBench | 0.0% | 0.0% |
| SWE-bench | 86.0% | 75.0% |
| Vibe Code Bench v1.1 | 64.8% | 20.2% |
| Agents | ||
| DTBench | 88.0% | 89.1% |
| LMCA | 41.4% | 43.1% |
| Professional work | ||
| Finance Agent (v2) | 48.5% | 42.5% |
| EMB | 59.7% | 26.0% |
| LegalBench | 82.4% | 86.9% |
| Legal Research Bench | 36.1% | 18.3% |
| MedCode | 28.7% | 55.9% |
| MedScribe | 83.9% | 69.9% |
| MortgageTax | 64.9% | 68.7% |
| Tax Agent Bench | 30.9% | 11.8% |
| TaxEval v2 | 70.9% | 73.9% |
| Novel problems | ||
| ARC-AGI-2 | 42.4% | 33.6% |
| ARC-AGI-1 | 87.5% | 84.7% |
Tests at least two of these models have taken. Bold rows are the core tests.