Which model leads τ³-Bench?

Q: Which model leads τ³-Bench?

MiMo V2.5 Pro tops the board at 72.9

τ³-Bench Leaderboard — Model Scores

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

The field · 18 measured

45.7872.9

18 models measured, most on the official card harness. MiMo V2.5 Pro tops the board at 72.9.

18 measured·Lifecycle Updated Oct 8, 2026Score basis

#◆Vendor$/M in$/M out

Leader — ignites whiteField — fill cools with scoreIndependently verified only0 of 18Bars show 45.78–72.9 · ◆ solid = first-party or better · outlined = arm's-length

01MiMo V2.5 ProXiaomi72.9$0.43/M$0.87/M$0.009

02GPT-5.4+1 altOpenAI72.9$2.5/M$15.0/M$0.120

03Claude Opus 4.6+1 altAnthropic72.4$5.0/M$25.0/M$0.207

04Qwen3.6 Plus+2 altsAlibaba70.7$0.50/M$3.0/M$0.025

05GLM-5.1+2 altsZ.ai70.6$1.4/M$4.4/M$0.041

06DeepSeek-V3.2+1 altDeepSeek69.2$0.28/M$0.42/M$0.005

07GLM-5+1 altZ.ai69.2$1.0/M$3.2/M$0.030

08Qwen3.5 35B A3BAlibaba68.93P$0.25/M$2.0/M$0.016

09Qwen3.5 27BAlibaba68.43P$0.30/M$2.4/M$0.020

10MiniMax M2.7+1 altMiniMax67.6$0.21/M$0.84/M$0.008

11Gemma 4 31BGoogle67.53P$0.17/M$0.40/M$0.004

12Qwen3.6 35B A3B+1 altAlibaba67.2$0.38/M$2.3/M$0.020

13Gemini 3.1 Pro+1 altGoogle67.1$2.0/M$12.0/M$0.104

14Kimi K2.5+1 altMoonshot66$0.45/M$2.3/M$0.020

15Granite 4.2 30B+1 altIBM62$0.16/M$0.65/M$0.007

16Gemma 4 26B A4BGoogle593P$0.07/M$0.34/M$0.003

17Granite 4.2 8B+2 altsIBM58.06$0.06/M$0.25/M$0.003

18Granite 4.2 3B+2 altsIBM45.78$0.03/M$0.12/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: 18 models scored · 0 independently verified · 8 vendor cross-reference · 10 vendor-reported · 0 with source disagreement. How these tiers are assigned