EnterpriseOps-Gym-AA
Completing business operations across eight domains in a ServiceNow environment.
How it is run
Artificial Analysis evaluates tool-using workflows and checks the final backend state. Highest published effort per model release is selected here.
What it cannot tell you
This ServiceNow implementation is a specific workflow environment; its success rate is separate from AutomationBench’s partial-credit score.
- Leader
- Claude Fable 551.12% · Artificial Analysis results, grouped here by release
- Top-ten spread
- 4.03%Leader minus tenth place, or last place in smaller fields.
- Median
- 40.56%of the 47 rows published
- No published result
- 24of 46 models on this site
Artificial Analysis results by release
Highest published effort per model release, grouped here from AA’s public results. Grouping ranks are derived here; source settings and model-publisher reports stay separate.
| # | Model | Score | Effort |
|---|---|---|---|
| 1 | Claude Fable 5 | 51.12% | max |
| 2 | Gemini 3.7 Flash | 50.40% | medium |
| 3 | Gemini 3.5 Flash | 50.07% | high |
| 4 | DeepSeek V4 Pro 0813 | 49.60% | max |
| 5 | Grok 4.6 | 48.34% | high |
| 6 | Qwen3.8 Max (0902) | 47.63% | Not specified |
| 7 | Claude Opus 5 | 47.48% | max |
| 8 | Qwen3.8 2.4T A95B | 47.36% | Not specified |
| 9 | Muse Spark 1.2 | 47.27% | xhigh |
| 10 | DeepSeek V4 Flash Vision | 47.09% | max |
| 11 | GPT-5.5 | 46.64% | xhigh |
| 12 | Kimi K3 | 45.33% | max |
| 13 | Qwen3.7 Max | 45.00% | Not specified |
| 14 | Claude Sonnet 5 | 44.67% | max |
| 15 | Qwen3.8 27B | 44.23% | xhigh |
| 16 | Claude Opus 4.8 | 43.96% | max |
| 17 | GPT-5.6 Sol | 42.91% | max |
| 18 | GLM-5.2 | 42.73% | max |
| 19 | Inkling Small | 42.70% | Not specified |
| 20 | Gemini 3.5 Flash-Lite | 42.35% | high |
| 21 | Gemini 3.1 Pro Preview | 42.17% | Not specified |
| 22 | GPT-5.6 Luna | 40.82% | max |
| 22 | Grok 4.5 | 40.82% | high |
| 24 | Qwen3.7 Plus | 40.56% | Not specified |
| 25 | DeepSeek V4 Pro 0424 | 40.44% | max |
| 26 | Kimi K2.7 Code | 40.23% | Not specified |
| 27 | DeepSeek V4 Flash 0420 | 39.57% | max |
| 28 | Kimi K2.6 | 38.53% | Not specified |
| 29 | GPT-5.6 Terra | 38.50% | max |
| 30 | Inkling | 37.96% | xhigh |
| 31 | GLM-5.3 | 36.44% | max |
| 32 | Muse Glimmer | 34.74% | high |
| 33 | GPT-5.4 mini | 34.62% | xhigh |
| 34 | Mistral Medium 3.5 | 33.72% | high |
| 35 | GLM 5.3 Flash | 33.21% | max |
| 36 | MiniMax-M3 | 32.11% | Not specified |
| 37 | GPT-5.4 nano | 31.48% | xhigh |
| 38 | Qwen3.5 397B A17B | 30.89% | Not specified |
| 39 | Claude 4.5 Haiku | 29.39% | Not specified |
| 40 | Step 3.7 Flash | 29.07% | high |
| 41 | Nemotron 3 Ultra 550B A55B | 28.89% | Not specified |
| 42 | Gemma 4 31B | 28.35% | Not specified |
| 43 | Gemini 3.1 Flash-Lite | 28.02% | high |
| 44 | Grok 4.3 | 26.26% | high |
| 45 | gpt-oss-120b | 25.54% | high |
| 46 | Nemotron 3 Super 120B A12B | 23.55% | Not specified |
| 47 | Nemotron 3.5 Lightning | 18.35% | Not specified |
No published result tracked
24 of the 46 models on this site have no published number on this benchmark, so their cells in the home table stay blank rather than reading as zero.
Sources
Every page used for this benchmark, read on 17 October 2026.
- Benchmark site or the page that publishes itartificialanalysis.ai/evaluations/enterpriseops-gym-aa
- Independent resultsartificialanalysis.ai/evaluations/enterpriseops-gym-aa