Which model leads CharadesSTA?

Q: Which model leads CharadesSTA?

Qwen3 VL 235B A22B Instruct tops the board at 64.8

CharadesSTA Leaderboard — Model Scores

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

The field · 11 measured

54.264.8

11 models measured, most on the official card harness. Qwen3 VL 235B A22B Instruct tops the board at 64.8.

11 measured·Lifecycle Updated Oct 17, 2026Score basis

#◆Vendor$/M in$/M out

Leader — ignites whiteField — fill cools with scoreIndependently verified only0 of 11Bars show 54.2–64.8 · ◆ solid = first-party or better · outlined = arm's-length

01Qwen3 VL 235B A22B InstructAlibaba64.8$0.21/M$1.4/M$0.012

02Qwen3 VL 235B A22B ReasoningAlibaba63.5$0.40/M$4.0/M$0.035

03Qwen3 VL 30B A3B InstructAlibaba63.5$0.20/M$0.80/M$0.008

04Qwen3 VL 32B ReasoningAlibaba62.8$0.16/M$0.64/M$0.006

05Qwen3 VL 30B A3B ReasoningAlibaba62.7$0.20/M$2.4/M$0.021

06Qwen3 VL 32B InstructAlibaba61.2$0.16/M$0.64/M$0.007

07Qwen3 VL Thinking (8B)Alibaba59.9$0.18/M$2.1/M$0.019

08Qwen3 VL 4B (Reasoning)Alibaba59———

09Qwen3 VL 8B InstructAlibaba56$0.18/M$0.70/M$0.008

10Qwen3 VL 4B InstructAlibaba55.5———

11Qwen2.5 VL 32B Instruct+1 altAlibaba54.2———

◆ 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: 11 models scored · 0 independently verified · 11 vendor-reported · 0 with source disagreement. How these tiers are assigned