Qwen3.8 27B vs Qwen3.8-Max
| Spec | Qwen3.8 27B | Qwen3.8-Max |
|---|---|---|
| Respan Index | 42.3 (#65) | 52.8 (#35) |
| Blended price per 1M | $0.154 | $2.48 |
| Context window | 262K | 1M |
| Input / output per 1M | $0.039 / $0.5 | $1.65 / $4.95 |
| Max output | - | 131K |
| Input | text, vision | text, vision |
| Released | 2026-08-14 | 2026-08-02 |
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 | Qwen3.8 27B | Qwen3.8-Max |
|---|---|---|
| Coding | 23.5 Qwen3.8 27B | 39.3 Qwen3.8-Max |
| Agents | 56.4 Qwen3.8 27B | 68.6 Qwen3.8-Max |
| Reasoning | 39.8 Qwen3.8 27B | 49.4 Qwen3.8-Max |
| Knowledge & language | 40.6 Qwen3.8 27B | 48.1 Qwen3.8-Max |
| Human preference | 55.6 Qwen3.8 27B | 58.0 Qwen3.8-Max |
Benchmarks both report
Qwen3.8-Max leads on 14 of the 16 benchmarks both report. Higher is better. Each score links to where it was published.
| Benchmark | Qwen3.8 27B | Qwen3.8-Max |
|---|---|---|
| LMArena Elo | 1435 | 1484 |
| LMArena Vision (LMArena) | 1244 | 1294 |
| LMArena WebDev (LMArena) | 1589 | 1671 |
| LMArena Agent (LMArena) | -0.055 | -0.0091 |
| APEX-Agents (Mercor) | 47.5% | 63.3% |
| LiveBench Agentic Coding (LiveBench) | 61.4% | 64.6% |
| LiveBench Coding (LiveBench) | 75.7% | 72.9% |
| LiveBench Data Analysis (LiveBench) | 76.6% | 78.4% |
| LiveBench Instruction Following (LiveBench) | 72.7% | 74.1% |
| LiveBench Language (LiveBench) | 74.3% | 79.7% |
| LiveBench Mathematics (LiveBench) | 86.2% | 91.3% |
| LiveBench Reasoning (LiveBench) | 80% | 88.2% |
| SAGE (Vals AI) | 52.4% | 51.3% |
| GPQA Diamond | 89.2% | 92.6% |
| SWE-Bench Pro | 61.7% | 67.7% |
| Terminal-Bench 2.1 | 73% | 86.6% |
FAQ
Is Qwen3.8 27B better than Qwen3.8-Max?
Qwen3.8-Max scores 52.8 on the Respan Index (#35), ahead of Qwen3.8 27B at 42.3 (#65). Qwen3.8-Max leads on 14 of the 16 benchmarks both report.
Which is cheaper, Qwen3.8 27B or Qwen3.8-Max?
Qwen3.8 27B lists at $0.039 input and $0.5 output per 1M tokens; Qwen3.8-Max at $1.65 and $4.95. Qwen3.8 27B is 16x cheaper per token at list price.
Which has the larger context window?
Qwen3.8 27B accepts 262K tokens and Qwen3.8-Max 1M tokens.
Compare other models
- vsQwen3.8 27B vs Qwen3.6 27B
- vsQwen3.8-Max vs Qwen3.7-Max
- vsClaude Opus 5.5 vs Qwen3.8-Max
- vsClaude Opus 5.5 vs Qwen3.8 27B
- vsQwen3.8 27B vs Muse Spark Open
- vsQwen3.8 27B vs Gemini 3 Flash Preview
- vsQwen3.8 27B vs Qwen3.7-Max
- vsQwen3.8-Max vs Kimi K3
- vsQwen3.8-Max vs DeepSeek V4.1 Flash
- vsQwen3.8-Max vs Grok 4.7