Qwen: Qwen3.5 397B A17B
Chatbailian/qwen3.5-397b-a17bQwen3.5-397B-A17B is the flagship model in Alibaba's Qwen3.5 series — the largest and most capable, featuring hybrid linear-attention and sparse MoE architecture. It leads the series in language understanding, logical reasoning, coding, and instruction following. Context: 256K tokens, output: 64K. Accessible via the OpenAI-compatible protocol.
Providers
/v1/chat/completionsCode Examples
from openai import OpenAIclient = OpenAI(base_url="https://api.ofox.io/v1",api_key="YOUR_OFOX_API_KEY",)response = client.chat.completions.create(model="bailian/qwen3.5-397b-a17b",messages=[{"role": "user", "content": "Hello!"}],)print(response.choices[0].message.content)
Benchmarks
Qwen: Qwen3.5 397B A17B scores 1442 in the Overall category of the LMArena text leaderboard (style control), ranking #65 of 374 models based on 54,977 human preference votes (updated 2026-07-12).
| Category | Arena Score | 95% CI | Rank | Votes |
|---|---|---|---|---|
| Overall | 1442 | 1438–1446 | #65 of 374 | 54,977 |
| Hard Prompts | 1463 | 1459–1468 | #59 of 374 | 35,438 |
| Coding | 1491 | 1486–1497 | #61 of 369 | 15,645 |
| Math | 1445 | 1435–1456 | #51 of 362 | 3,188 |
| Creative Writing | 1408 | 1401–1415 | #67 of 372 | 8,556 |
| Instruction Following | 1433 | 1428–1439 | #62 of 374 | 17,752 |
| Chinese | 1502 | 1490–1514 | #34 of 344 | 2,710 |
Source: LMArena · CC BY 4.0 · Updated 2026-07-12 · Methodology ↗ · Ranks compare models within each category of the LMArena text leaderboard (style control). Scores come from third-party human preference evaluations, not from OFOX.