From 1c1db172152f32be2428b5643dcead96d2ec8e0e Mon Sep 17 00:00:00 2001 From: Linghong Date: Wed, 8 Jul 2026 14:39:44 +0000 Subject: [PATCH] =?UTF-8?q?=E8=83=8C=E6=99=AF=E8=89=B2=E5=86=B3=E7=AD=96?= =?UTF-8?q?=20max=5Ftokens=20768=20=E2=86=92=201024?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 给 gpt-5-mini low 推理档再留余量（实测推理 320~384 token）， 降低复杂图触顶后走空响应兜底的概率。 Co-Authored-By: Claude Opus 4.8 --- .../api-server/src/editor_screen_background_decision.rs | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/server-rs/crates/api-server/src/editor_screen_background_decision.rs b/server-rs/crates/api-server/src/editor_screen_background_decision.rs index 1762e3e92..48c47c27f 100644 --- a/server-rs/crates/api-server/src/editor_screen_background_decision.rs +++ b/server-rs/crates/api-server/src/editor_screen_background_decision.rs @@ -152,9 +152,9 @@ pub(crate) async fn resolve_editor_screen_background_color( None => LlmMessage::user(user_prompt.as_str()), }; // 预算要够推理模型（如 gpt-5-mini）先花几百 token 推理、再吐 JSON 答案； - // 实测 low 档推理约 320~384 token，取 768 留足余量。非推理模型遇 stop 提前结束，不会多花。 + // 实测 low 档推理约 320~384 token，取 1024 留足余量。非推理模型遇 stop 提前结束，不会多花。 let mut request = LlmTextRequest::new(vec![LlmMessage::system(system_prompt), user_message]) - .with_max_tokens(768) + .with_max_tokens(1024) .with_request_timeout_ms(EDITOR_SCREEN_BACKGROUND_DECISION_TIMEOUT_MS); if let Some(vision_model) = vision_model { // 视觉模型 gpt-5-mini 是推理模型：走 Responses 协议并压到 low 推理档，