Newsbite analysis · AI, Business
Dev.to
Mistral Large 4 Scores 59.9% on AutomationBench Across Four Workplace Apps
Mistral AI has introduced its Mistral Large 4 (ML4) model, which has achieved a score of 59.9% on AutomationBench across 657 business workflows, including simulated environments for Gmail, Google Sheets, Slack, and Salesforce. This model aims to enhance cross-application automation, addressing how workplaces coordinate tasks across various digital tools. ML4 outperforms competitors like Kimi K3 and DeepSeek V4 Pro, although its benchmark score indicates potential rather than definitive performance in live environments.
MonitorConfidence: HighMid term horizonStory: First report
EvidenceOfficial announcementIndependent evidenceNo quotes
Published 7 Oct 2026, 16:51 UTC1 min read
Lead image — Dev.to
Brief
NewsBite reading:Mistral Large 4 Achieves 59.9% on AutomationBench in Workflow Tests
TL;DR
- Mistral Large 4 scored 59.9% across 657 business workflows.
- ML4 was tested in simulated Gmail, Google Sheets, Slack, and Salesforce environments.
- The model surpassed Kimi K3, MiMo-V2.6-Pro, and DeepSeek V4 Pro in AutomationBench results.
- Mistral plans to release the model's open weights by October 2026.
- ML4 aims to improve cross-application workflow automation for enterprises.
Interesting facts
- The score is a benchmark, not a guarantee of performance in real business processes.
- AutomationBench focuses on cross-application orchestration rather than isolated tasks.
- Mistral is positioning ML4 as a foundation model for future specialized successors.