Which model leads VideoMME (w sub.)?

Q: Which model leads VideoMME (w sub.)?

Qwen2.5 VL 72B tops the board at 79.1

VideoMME (w sub.) Leaderboard — Model Scores

Pro Mode — raw scores, one benchmark under the lensthird-party eval harness

The field · 7 measured

62.179.1

7 models measured, most on the third-party eval harness. Qwen2.5 VL 72B tops the board at 79.1.

7 measured·Lifecycle Updated Oct 8, 2026Score basis

#◆Vendor$/M in$/M out

Leader — ignites whiteField — fill cools with scoreIndependently verified only0 of 7Bars show 62.1–79.1 · ◆ solid = first-party or better · outlined = arm's-length

01Qwen2.5 VL 72BAlibaba79.1———

02Qwen2.5 VL 32B InstructAlibaba77.9———

03GPT-4o+1 altOpenAI77.2$5.0/M$15.0/M$0.130

04Kimi VL A3B Instruct+1 altMoonshot72.6VENDOR———

05GPT-4o miniOpenAI68.9$0.15/M$0.60/M$0.005

06Kimi VL A3B Thinking+1 altMoonshot66———

07Gemma 3 12B+1 altGoogle62.1$0.05/M$0.15/M$0.002

◆ 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: 7 models scored · 0 independently verified · 5 vendor cross-reference · 2 vendor-reported · 1 with source disagreement. How these tiers are assigned