fix(llm): 改用 max_tokens 避免与 ST 代理注入的 max_tokens 冲突

ST 的 /api/backends/chat-completions/generate 端点会从预设配置注入
max_tokens,导致与扩展自行传递的 max_completion_tokens 冲突,
上游 API 返回 400。改为 max_tokens 后由 ST 统一处理,消除冲突。

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
Hao19911125
2026-04-08 16:08:57 +08:00
parent 20f64138b1
commit 4d2f97397c

View File

@@ -1441,7 +1441,7 @@ async function callDedicatedOpenAICompatible(
model: config.model,
messages,
temperature: filteredGeneration.temperature ?? 1,
max_completion_tokens: resolvedCompletionTokens,
max_tokens: resolvedCompletionTokens,
stream: filteredGeneration.stream ?? false,
frequency_penalty: filteredGeneration.frequency_penalty ?? 0,
presence_penalty: filteredGeneration.presence_penalty ?? 0,