Qwen3.6 27B vs Qwen3.8 27B
| Spec | Qwen3.6 27B | Qwen3.8 27B |
|---|---|---|
| Respan Index | 22.3 (#116) | 42.3 (#65) |
| Blended price per 1M | $0.688 | $0.154 |
| Context window | 262K | 262K |
| Input / output per 1M | $0.25 / $2 | $0.039 / $0.5 |
| Max output | - | - |
| Input | text, vision | text, vision |
| Released | 2026-04-21 | 2026-08-14 |
Respan Index by category
Each category is scored 0 to 100, and how far ahead the winner is. Leads under a point are close to even.
| Category | Qwen3.6 27B | Qwen3.8 27B |
|---|---|---|
| Coding | 6.2 Qwen3.6 27B | 23.5 Qwen3.8 27B |
| Agents | 22.3* Qwen3.6 27B | 56.4 Qwen3.8 27B |
| Reasoning | 26.1 Qwen3.6 27B | 39.8 Qwen3.8 27B |
| Knowledge & language | 17.9 Qwen3.6 27B | 40.6 Qwen3.8 27B |
| Human preference | 45.7* Qwen3.6 27B | 55.6 Qwen3.8 27B |
* Estimated: no published scores in that category. How the Respan Index works
Benchmarks both report
Qwen3.8 27B leads on 9 of the 9 benchmarks both report. Higher is better. Each score links to where it was published.
| Benchmark | Qwen3.6 27B | Qwen3.8 27B |
|---|---|---|
| GPQA Diamond | 87.8% | 89.2% |
| LiveBench Agentic Coding (LiveBench) | 39.3% | 61.4% |
| LiveBench Coding (LiveBench) | 71.8% | 75.7% |
| LiveBench Data Analysis (LiveBench) | 70.4% | 76.6% |
| LiveBench Instruction Following (LiveBench) | 53.2% | 72.7% |
| LiveBench Language (LiveBench) | 63.3% | 74.3% |
| LiveBench Mathematics (LiveBench) | 79.9% | 86.2% |
| LiveBench Reasoning (LiveBench) | 70.3% | 80% |
| SAGE (Vals AI) | 45.6% | 52.4% |
FAQ
Is Qwen3.6 27B better than Qwen3.8 27B?
Qwen3.8 27B scores 42.3 on the Respan Index (#65), ahead of Qwen3.6 27B at 22.3 (#116). Qwen3.8 27B leads on 9 of the 9 benchmarks both report.
Which is cheaper, Qwen3.6 27B or Qwen3.8 27B?
Qwen3.6 27B lists at $0.25 input and $2 output per 1M tokens; Qwen3.8 27B at $0.039 and $0.5. Qwen3.8 27B is 4.5x cheaper per token at list price.
Which has the larger context window?
Both accept 262K tokens of context.