ros/claude-task-master
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: improve config-manager max tokens for openrouter and kimi-k2 model (#1035)

Browse Source
This commit is contained in:
Ralph Khreish
2025-07-23 19:03:26 +03:00
committed by GitHub
parent bdb11fb2db
commit fb7d588137
3 changed files with 27 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
.changeset/quiet-rabbits-bathe.md Normal file
View File

@@ -0,0 +1,10 @@
---
"task-master-ai": patch
---
Fix max_tokens limits for OpenRouter and Groq models
- Add special handling in config-manager.js for custom OpenRouter models to use a conservative default of 32,768 max_tokens
- Update qwen/qwen-turbo model max_tokens from 1,000,000 to 32,768 to match OpenRouter's actual limits
- Fix moonshotai/kimi-k2-instruct max_tokens to 16,384 to match Groq's actual limit (fixes #1028)
- This prevents "maximum context length exceeded" errors when using OpenRouter models not in our supported models list