RobustBench-TC - Response-Time Description (Abbreviated Name): leaderboard

RobustBench-TC (2026) tests tool-calling models under deployment noise grounded in real tool-calling failures, using 199 single-turn samples drawn from BFCL, API-Bank, RoTBench, ToolAlpaca and ToolEyes and scored by deterministic rule-based parsers. This board is tool-call accuracy in percent under response-time annotations in the tool descriptions and an abbreviated name on the correct tool; higher is better.

Metric: Tool-call accuracy (%) under the TimeDesc-Abbr perturbation (reward component: TimeDesc + abbreviated GT name); RobustBench-TC samples drawn from BFCL V3 single-turn, API-Bank, RoTBench, ToolAlpaca and ToolEyes, each scored by its source benchmark's strict scorer after a format-tolerant tool-call parse, temperature 0 (DeepSeek-R1-Distill at 0.6), Qwen3-family models with thinking disabled; higher is better. Source: arxiv.org. Saturation forecast: Around 2031. 18 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Qwen 3 32BNon-reasoning21.5#983
2Qwen 3 14BNon-reasoning18.8#1284
3MUA-RL-32B18.8
4MUA-RL-8B17.4#1656
5MUA-RL-14B17.4#1674
6LoopTool-32B16
7Qwen 3 8BNon-reasoning15.3#1890
8LoopTool-8B14.6#1497
9DeepSeek R1 Distill Qwen 14B13.2#2739
10Qwen 3.5 9BNon-reasoning11.8#838
11Llama 3.2 3B Instruct11.1#7311
12O4 Mini11.1#372
13ToolRL-Llama3.2-3B11.1
14ToolRL-Qwen2.5-1.5B9#2330
15Qwen 2.5 1.5B Instruct7.6#7477

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=robustbench-tc-response-time-description-abbreviated-name · How It Works · Data refreshed daily, snapshot 2026-11-22.