1.15.27 1.15.28

Name: crewAI
Author: crewAIInc

low risk

crewAIInc/crewAIOct 9, 2026 → Oct 14, 20264 days between releases

What's New

Adds new crewai eval command and LiteLLM catalog integration

New Features

Add crewai eval command to report where a stopped model broke
Use LiteLLM catalog for context windows

Improvements

Improved error reporting for deployment failures and lockfile checks
Refresh LiteLLM catalog after the version-bump PR merges
Decode local JSON files with UTF-8 BOM in rag

Bug Fixes

Fix flow to learn from resumed human feedback
Fix trigger list and payload requests timeout to 60 seconds
Read AMP's source_type during deploy push
Stop JSON-schema models from warning about BaseModel-shadowing fields
Preserve native tool aborts and post-hook cache identity

Upgrade Notes

New `crewai eval` command is available for debugging model failures. LiteLLM catalog is now integrated for context window management. Monitor for warnings regarding zero platform tools.

The Two Releases

1.15.27

Oct 9, 2026· 12 days ago

This release expands provider support by adding XPU and DeepInfra options, improves evaluation tracking with cost and time metrics, and fixes critical bugs regarding OpenAI model parameters and SQLite initialization.

Full release notes

1.15.28

Oct 14, 2026· 8 days ago

This release introduces a new evaluation command to debug model failures and integrates the LiteLLM catalog for better context window management. It also includes several bug fixes concerning flows, deployment processes, and tool resolution.

Full release notes