Qwen3.5 397B A17B
Open Sourcevon Alibaba Qwen · Veroeffentlicht 2026-02-16
78.4
Durchschn. Score
$0.39/1M
Eingabepreis
$2.34/1M
Ausgabepreis
262K tokens (~131 books)
Kontextfenster
multimodal
Typ
Tested on 11 benchmarks with 78.4% average. Top scores: Chatbot Arena Elo — Overall (1447.7%), Chatbot Arena Elo — Coding (1386.1%), OpenCompass — AIME2025 (92.3%).
Benchmark-Ergebnisse
| Benchmark | Kategorie | Score | Bar |
|---|---|---|---|
| Chatbot Arena Elo — Overall | arena | 1447.7 | |
| Chatbot Arena Elo — Coding | arena | 1386.1 | |
| OpenCompass — AIME2025 | math | 92.3 | |
| OpenCompass — IFEval | language | 91.5 | |
| OpenCompass — GPQA-Diamond | knowledge | 88.4 | |
| OpenCompass — MMLU-Pro | knowledge | 87.6 | |
| OpenCompass — LiveCodeBenchV6 | coding | 83.0 | |
| Artificial Analysis — Agentic Index | speed | 55.8 | |
| Artificial Analysis — Quality Index | speed | 45.0 | |
| Artificial Analysis — Coding Index | speed | 41.3 | |
| OpenCompass — HLE | knowledge | 27.5 |
Aehnliche Modelle
DeepSeek
78.2
Anthropic
78.0
stepfun
76.9
Google DeepMind
76.9