List your company

Tool Use on τ-Bench (TauB) V2 (accuracy)

91.6AccuracyQwen3.5-122B
7.25629.15351.0572.947May 16, 2026

Evaluation results

MethodLinks
Qwen3.5-122BDeployment=Local model2026.0591.6
Best LocalConfig=Oracle local routing frontier2026.0591.6
Gemma4-26BDeployment=Local model2026.0591.3
Gemini 3.1 ProDeployment=Cloud baseline2026.0590.8
Qwen3.5-35BDeployment=Local model2026.0590.2
Claude Opus 4.6Deployment=Cloud baseline2026.0589.5
GPT 5.4Deployment=Cloud baseline2026.0589.2
Qwen3.5-27BDeployment=Local model2026.0588.4
Qwen3.5-9BDeployment=Local model2026.0577.1
Gemma4-E4BDeployment=Local model2026.0556.1
Nemotron-Super-120BDeployment=Local model2026.0536.8
Granite 4.0 H-SmallDeployment=Local model2026.0517.5
Granite 3.3 8BDeployment=Local model2026.0510.5