Inference
EN

AutomationBench

An agent benchmark for completing automation workflows in reproducible task environments.

ModelsModels with a result.32
Top resultBest result on this benchmark.62.2%MiniMax M3.1 Flash
Top-3 spreadPoints from first to third.1.7 pts
YearYear of release.—

Result and price

29 of 32 results

Results

32 results
#ModelResultThe score from the source.
01MiniMax M3.1 FlashAbove fitFar above this model's index.62.2%
02ERNIE 5.5 Flash Preview61.9%
03DeepSeek V4.1 Flash 1017Above fitFar above this model's index.60.5%
04Hy4Above fitFar above this model's index.56.2%
05Gemini 4 Carbon53.9%
06Atria Dawn Preview53.8%
07MiMo-V2.6-ProAbove fitFar above this model's index.53.1%
08Ling 3.1 FlashAbove fitFar above this model's index.52.5%
09MiMo-V2.6-FlashAbove fitFar above this model's index.52.3%
10LongCat-2.551.8%
11Claude Fable 5.251.4%
12Gemini 4 Argon51.3%
13Ultramarine 1 Flash50.2%
14Muse Spark 1.349.4%
15GLM-5.3-Flash48.8%
16GLM-5.348.2%
17Step 5 Preview44.0%
18GPT-6 Astra41.4%
19LFM3-1.7BAbove fitFar above this model's index.41.3%
20Claude Opus 5.540.0%
21Ripple-65B Preview36.4%
22GPT-6.1 Sol36.1%
23GPT-6 Sol33.2%
24Hy4 preview32.1%
25DeepSeek V4 Pro 081331.8%
26Claude Fable 5.131.4%
27Kimi K330.8%
28Gemini 3.7 Flash30.4%
29LFM2.5-24B-A2B27.7%
30Qwen3.8 Max27.3%
31Claude Opus 526.0%
32DeepSeek V4 Flash 073125.1%