Up to four models
Qwen3.8-27B vs Qwen3.6 27B
On the index (GPT-4 = 100), Qwen3.8-27B scores 161 and Qwen3.6 27B 147. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
Qwen3.8-27B161145–178
WeightsOpen
ReleasedAug 2026
Price in / out$0.039 / $1.25
Context1M
Core tests taken3 of 9
Qwen3.6 27B147130–163
WeightsOpen
ReleasedApr 2026
Price in / out$0.32 / $3.2
Context262K
Core tests taken2 of 9
Qwen3.8-27B leads Qwen3.6 27B by 15 points, and costs 3.0× less.
Test by testthird-party results
Qwen3.8-27BQwen3.6 27BEvery result as a table
| Test | Qwen3.8-27B | Qwen3.6 27B |
|---|---|---|
| Reasoning & knowledge | ||
| LiveBench Data Analysis | 76.6% | 70.4% |
| LiveBench IF | 72.7% | 53.2% |
| LiveBench Language | 74.4% | 63.3% |
| LiveBench Reasoning | 80.0% | 70.3% |
| Maths | ||
| LiveBench Mathematics | 86.2% | 79.9% |
| Coding | ||
| LiveBench Agentic Coding | 61.4% | 39.3% |
| LiveBench Coding | 75.7% | 71.8% |
| SWE-bench | 86.0% | 70.0% |
| Vibe Code Bench v1.1 | 64.8% | 11.9% |
| Agents | ||
| DTBench | 88.0% | 78.1% |
| LMCA | 41.4% | 34.5% |
| Professional work | ||
| MortgageTax | 64.9% | 68.3% |
| SAGE | 52.4% | 45.6% |
| TaxEval v2 | 70.9% | 71.3% |
Tests at least two of these models have taken. Bold rows are the core tests.
