Which model leads VITA-Bench?

Q: Which model leads VITA-Bench?

Qwen3.5 397B A17B tops the board at 49.7

VITA-Bench Leaderboard — Model Scores

Pro Mode — raw scores, one benchmark under the lensofficial card harness

The field · 13 measured

2249.7

13 models measured, most on the official card harness. Qwen3.5 397B A17B tops the board at 49.7.

13 measured·Lifecycle Updated Oct 8, 2026Score basis

#◆Vendor$/M in$/M out

Leader — ignites whiteField — fill cools with scoreIndependently verified only0 of 13Bars show 22–49.7 · ◆ solid = first-party or better · outlined = arm's-length

01Qwen3.5 397B A17BAlibaba49.7$0.60/M$3.6/M$0.042

02Qwen3.7 MaxAlibaba47.9$2.5/M$7.5/M$0.104

03Qwen3.7 Plus PreviewAlibaba45.6$0.40/M$1.6/M$0.022

04Qwen3.6 PlusAlibaba44.3$0.50/M$3.0/M$0.040

05Gemma 4 31BGoogle433P$0.17/M$0.40/M$0.007

06Qwen3.5 27B+1 altAlibaba41.9$0.30/M$2.4/M$0.032

07Qwen3 Max (Reasoning)Alibaba40.9$1.2/M$6.0/M$0.088

08Gemma 4 26B A4BGoogle36.93P$0.07/M$0.34/M$0.006

09Qwen3.6 35B A3B+1 altAlibaba35.6$0.38/M$2.3/M$0.037

10Qwen3.5 122B A10BAlibaba33.6$0.40/M$3.2/M$0.054

11Qwen3.5 35B A3B+1 altAlibaba31.9$0.25/M$2.0/M$0.035

12Qwen3.5 9BAlibaba29.8$0.14/M$0.20/M$0.006

13Qwen3.5 4BAlibaba22$0.03/M$0.15/M$0.004

◆ solid = independently verified · aggregator attested · vendor attributed — outlined = cross-reference · source-attributed. Dimmed ◆ = confidence inferred. “+N alts” = alternate disclosures in the drawer with why-demoted. Prices per million tokens; $/point = input price ÷ score. Every row opens its model dossier.

Verification: 13 models scored · 0 independently verified · 2 vendor cross-reference · 11 vendor-reported · 0 with source disagreement. How these tiers are assigned