Description

This PR addresses a specific compatibility issue between the Continue OpenAI adapter and DeepSeek Reasoner (R1) models.

The Problem

1. Mandatory Field: DeepSeek's API is strict regarding the structure of assistant messages in the chat history. When a previous message contains tool_calls, the API expects a reasoning_content field to be present (even if empty) in the subsequent request. Missing this field results in a 400 Bad Request error.
2. Token Mapping: Similar to OpenAI's o-series, DeepSeek Reasoner models utilize max_completion_tokens instead of the standard max_tokens for managing reasoning and output budgets.

The Solution

• Modified modifyChatBody in OpenAIApi.ts to identify official DeepSeek endpoints or reasoner models.
• Payload Transformation: Added a mapping layer that ensures reasoning_content: "" is added to assistant messages that contain tool calls but lack reasoning data.
• Parameter Mapping: Automatically renames max_tokens to max_completion_tokens for DeepSeek reasoner models to ensure request validity.

Fixes #10475

AI Code Review

• Team members only: AI review runs automatically when PR is opened or marked ready for review
• Team members can also trigger a review by commenting @continue-review

Checklist

• I've read the contributing guide
• The relevant docs, if any, have been updated or created
• The relevant tests, if any, have been updated or created

Screen recording or screenshot

[ When applicable, please include a short screen recording or screenshot - this makes it much easier for us as contributors to review and understand your changes. See this PR as a good example. ]

Tests

• Logic Verification: Fix is based on DeepSeek's official API schema requirements to resolve the "Missing reasoning_content" 400 error during tool-use history.
• Payload Inspection: Confirmed via local builds that modifyChatBody correctly injects reasoning_content and maps token parameters specifically for DeepSeek.
• Regression Check: Verified that the changes are isolated to DeepSeek endpoints, ensuring zero impact on standard OpenAI or other provider logic.
• Build Status: npm run build passed successfully in packages/openai-adapters.

Summary by cubic

Fixes DeepSeek Reasoner (R1) compatibility in the OpenAI adapter by adding missing reasoning_content and mapping max_tokens to max_completion_tokens to prevent 400 errors.

• Bug Fixes
  • For DeepSeek endpoints or models, add reasoning_content: "" to assistant messages that have tool_calls but no reasoning data.
  • Map max_tokens to max_completion_tokens for DeepSeek reasoner models; other providers are untouched.

^{Written for commit 62b4e77. Summary will update on new commits.}