[PR #190] fix(pi): migrate to pi's current model config format #258

Open
opened 2026-10-26 09:47:20 +00:00 by bsakaguchi · 0 comments
Owner

📋 Pull Request Information

Original PR: https://github.com/deepseek-ai/awesome-deepseek-agent/pull/190
Author: @0xlxx
Created: 6/5/2026
Status: 🔄 Open

Base: main ← Head: main


📝 Commits (1)

  • d98b292 fix(pi): update models.json to current pi format

📊 Changes

2 files changed (+40 additions, -36 deletions)

View changed files

📝 docs/pi_mono.md (+20 -18)
📝 docs/pi_mono.zh-CN.md (+20 -18)

📄 Description

Summary

Pi deprecated compat.reasoningEffortMap in favor of model-level thinkingLevelMap. This PR updates the Pi integration guides (EN + zh-CN) to use the current format.

Changes

  • Replace deprecated compat.reasoningEffortMap with thinkingLevelMap
  • Add supportsDeveloperRole: false at provider level (DeepSeek rejects the developer role)
  • Move thinkingFormat from model-level compat to provider-level compat (reduces duplication)

Verification

Per pi's models documentation:

Migration: older configs that used compat.reasoningEffortMap should move that mapping to model-level thinkingLevelMap.

The supportsDeveloperRole field is documented under OpenAI Compatibility — DeepSeek requires this to be false to avoid 400 errors from the developer role.


🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/deepseek-ai/awesome-deepseek-agent/pull/190 **Author:** [@0xlxx](https://github.com/0xlxx) **Created:** 6/5/2026 **Status:** 🔄 Open **Base:** `main` ← **Head:** `main` --- ### 📝 Commits (1) - [`d98b292`](https://github.com/deepseek-ai/awesome-deepseek-agent/commit/d98b292f44071aeb71c86ee1d1f967fdf3cf2e2f) fix(pi): update models.json to current pi format ### 📊 Changes **2 files changed** (+40 additions, -36 deletions) <details> <summary>View changed files</summary> 📝 `docs/pi_mono.md` (+20 -18) 📝 `docs/pi_mono.zh-CN.md` (+20 -18) </details> ### 📄 Description ## Summary Pi deprecated `compat.reasoningEffortMap` in favor of model-level `thinkingLevelMap`. This PR updates the Pi integration guides (EN + zh-CN) to use the current format. ## Changes - Replace deprecated `compat.reasoningEffortMap` with `thinkingLevelMap` - Add `supportsDeveloperRole: false` at provider level (DeepSeek rejects the `developer` role) - Move `thinkingFormat` from model-level compat to provider-level compat (reduces duplication) ## Verification Per pi's [models documentation](https://github.com/badlogic/pi-mono/tree/main/packages/coding-agent/docs/models.md): > Migration: older configs that used `compat.reasoningEffortMap` should move that mapping to model-level `thinkingLevelMap`. The `supportsDeveloperRole` field is documented under [OpenAI Compatibility](https://github.com/badlogic/pi-mono/tree/main/packages/coding-agent/docs/models.md#openai-compatibility) — DeepSeek requires this to be `false` to avoid 400 errors from the `developer` role. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
Sign in to join this conversation.
No labels
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
deepseek-ai/awesome-deepseek-agent#258
No description provided.