AutomationBench

Fresh

Evaluates AI agents on realistic, multi-step business workflows across 47 simulated SaaS tools.

Type: RL Env
Publisher: Zapier
Tags: Tool Use Business Workflows
Runtime: tool-use
License: unknown
Size: v1.0.7
Published: Apr 2026
Updated: Sep 2026
Canonical: app.primeintellect.ai/dashboard/environments/zapier/AutomationBench

Cite

Notes

Only stored in your browser.

Attribution

README: api.primeintellect.ai/api/v1/environmentshub/zapier/AutomationBench/@1.0.5/inspect
Scores: prime-hubAnthropicOpenAI

Attribution policy →

Public scores on this env

17

18 vf-eval reports across 17 models

1Claude Opus 4.6Anthropic56.1%2Claude Sonnet 4.6Anthropic47.6%3GPT-6 AstraOpenAI41.4%4MiMo-V2.5-ProXiaomi39.4%5GPT-5.4 MiniOpenAI37.0%6GLM 5.1Zai35.9%7GPT-6 SolOpenAI33.2%8Gemini 3 Flash PreviewGoogle (Alphabet Inc.)32.7%9MiMo-V2.5Xiaomi32.5%10GPT-5.4 NanoOpenAI26.2%11Nemotron 3 Super 120B A12BNVIDIA20.7%12Claude Fable 5Anthropic17.4%13Claude Opus 4.8Anthropic15.5%14GPT-5.4OpenAI13.1%15GPT-5.5OpenAI12.9%16Gemini 3.1 Pro PreviewGoogle (Alphabet Inc.)12.3%17MiniMax M2.7Minimax5.7%

Open the scoring view →