Which model leads tau2-bench telecom?

Q: Which model leads tau2-bench telecom?

DeepSeek-V4-Flash tops the board at 98.2

tau2-bench telecom Leaderboard — Model Scores

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

The field · 6 measured

73.1998.2

6 models measured, most on the third-party eval harness. DeepSeek-V4-Flash tops the board at 98.2.

6 measured·6 new this week·Lifecycle Updated Oct 17, 2026Score basis

#◆Vendor$/M in$/M out

Leader — ignites whiteField — fill cools with scoreIndependently verified only0 of 6Bars show 73.19–98.2 · ◆ solid = first-party or better · outlined = arm's-length

01DeepSeek-V4-FlashNEWDeepSeek98.2$0.01/M$0.30/M$0.002

02Qwen3.8 Flash NextNEWAlibaba90.4$0.15/M$0.47/M$0.003

03Inkling-SmallNEWThinking Machines90.1$0.30/M$1.2/M$0.008

04Qwen3.5 122B A10BNEWAlibaba87.2$0.40/M$3.2/M$0.021

05Nemotron 3 Ultra 550B A55BNEWNVIDIA83$0.60/M$2.7/M$0.020

06NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16NEWNVIDIA73.19———

◆ 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: 6 models scored · 0 independently verified · 4 vendor cross-reference · 2 vendor-reported · 0 with source disagreement. How these tiers are assigned