Qwen3.5 397B A17B
Qwen 3.5's open-source 397B MoE model (17B active params) with hybrid linear attention. Supports text, image, and video input with a 256K context window.
- Vision
- Video Input
Added Feb 16, 2026
Model weightsPricing
Auto routing · per 1M tokens- Input
- $0.495
- Output
- $3.30
- Cache read
- $0.242
Specifications
- Context window
- 262.1K
- Max output
- 65.5K
- Parameters
- 397B / 17B
- Total / active
Benchmarks
Benchmarks
Sourced from LMArena.
Arena Score
1441.5
Overall Rank
#96 / 421
Votes
97,678
Confidence Interval
1438.4 - 1444.5
Category Scores
Coding
#96 / 416
27,511 votes
1490.4
Math
#79 / 404
5,360 votes
1450.1
Longer Query
#85 / 399
42,910 votes
1453.6
Creative Writing
#102 / 419
17,467 votes
1406.0
Instruction Following
#91 / 421
33,482 votes
1434.5
Hard Prompts
#92 / 421
64,306 votes
1463.4
Additional Categories22
Chinese
#66 / 395
6,867 votes
1496.5
Japanese
#66 / 288
1,221 votes
1427.6
French
#76 / 301
3,184 votes
1471.5
Expert
#82 / 371
10,369 votes
1477.8
Industry Medicine And Healthcare
#84 / 390
7,265 votes
1466.6
Industry Business And Management And Financial Operations
#85 / 414
19,367 votes
1446.9
Industry Life And Physical And Social Science
#86 / 419
15,943 votes
1465.6
Industry Mathematical
#86 / 401
5,645 votes
1452.9
Spanish
#87 / 303
3,141 votes
1442.5
Multi Turn
#88 / 419
15,745 votes
1452.0
Polish
#88 / 244
1,757 votes
1446.5
German
#89 / 320
1,536 votes
1442.3
Non English
#89 / 421
55,027 votes
1432.3
Korean
#95 / 289
1,578 votes
1391.7
Industry Writing And Literature And Language
#96 / 420
24,257 votes
1420.7
Industry Software And It Services
#97 / 421
38,833 votes
1479.9
Exclude Ties
#98 / 421
71,526 votes
1440.4
Russian
#98 / 385
11,341 votes
1437.4
Hard Prompts English
#100 / 419
28,631 votes
1466.4
English
#101 / 421
42,646 votes
1448.5
Industry Legal And Government
#101 / 393
7,474 votes
1447.4
Industry Entertainment And Sports And Media
#107 / 419
22,212 votes
1402.6
Published 2026-11-09 · Matched as Qwen3.5-397B-A17B · CC BY 4.0
LMArena DatasetLeaderboard results
| Benchmark | Score | Rank on NanoGPT |
|---|---|---|
Epoch AI Benchmarking Hub Epoch AI data, CC BY 4.0 · Updated Nov 12, 2026 | ||
Epoch Capabilities Index | 147.9 | #69 of 115 |
FrontierMath v2 (Tiers 1–3) | 31.2% | #68 of 83 |
GPQA Diamond | 86.4% | #62 of 119 |
OTIS Mock AIME 2024–2025 | 88.9% | #62 of 121 |
Mystery Game Puzzles | 18.0% | #55 of 77 |
Sierra τ-bench tau2-bench data, MIT · Updated Aug 4, 2026 | ||
τ²-bench Overall | 87.9% | #1 of 6 |
Creative Writing v3 | 1478 | #59 of 96 |
Longform Writing | 68.3 | #43 of 97 |
EQ-Bench 3 | 1451 | #25 of 49 |
UGI Leaderboard UGI Leaderboard data, Apache-2.0 · Updated Oct 16, 2026 | ||
UGI Score | 28.5 | #110 of 149 |
UGI Willingness | 1.8 | #138 of 149 |
UGI Writing | 50.1 | #59 of 149 |
Filtered and ranked to models on NanoGPT.
Providers
Choose or restrict providers with the API. Provider-selection prices include the selection markup. Provider selection guide · Pricing API (JSON, USD per 1K tokens)
 | $0.495 | $3.30 | $0.242 | N/A | N/A |
| $0.495 | $3.30 | $0.242 | N/A | N/A |
| $0.41 | $2.46 | N/A | N/A | N/A | |
| $0.472 | $3.15 | $0.231 | N/A | N/A | |
| $0.525 | $3.78 | $0.315 | N/A | N/A | |
| $0.578 | $3.68 | $0.578 | N/A | N/A | |
| $0.579 | $3.78 | N/A | N/A | N/A | |
| $0.63 | $3.78 | N/A | N/A | N/A | |
| $0.63 | $3.78 | N/A | N/A | N/A | |
| $0.63 | $3.78 | N/A | N/A | N/A | |
| $0.63 | $3.78 | $0.126 | N/A | N/A | |
| $0.704 | $4.20 | N/A | N/A | N/A | |
| $0.787 | $4.73 | N/A | N/A | N/A |
Batch rows for this model use normal routing and cost the same as standard API requests. Set provider in a row to choose a provider.