From 49ebe4bd6197003cc5dec97c1989a4b2a70601c8 Mon Sep 17 00:00:00 2001
From: Vimsucks <dev@vimsucks.com>
Date: Sun, 5 Apr 2026 03:22:04 +0800
Subject: [PATCH] Add reasoning_effort field to OpenAI compatible model
 configuration (#50582)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Some model like glm-5、kimi-k2.5 support reasoning, but require
reasoning_effort parameter

This pr add support for setting reasoing_effort for openai compatible
models

Tested using the following config:

```json
{
  "language_models": {
    "openai_compatible": {
      "My LiteLLM": {
        "available_models": [
          {
            "name": "glm-5",
            "display_name": "glm-5",
            "max_tokens": 73728,
            "reasoning_effort": "low"
          },
          {
            "name": "kimi-k2.5",
            "display_name": "kimi-k2.5",
            "max_tokens": 262144,
            "reasoning_effort": "low"
          }
        ]
      }
    }
  }
}

```

Release Notes:

- Added a setting to control `reasoning_effort` in custom
OpenAI-compatible models
---
 .../src/agent_configuration/add_llm_provider_modal.rs         | 1 +
 crates/language_models/src/provider/open_ai_compatible.rs     | 4 ++--
 crates/settings_content/src/language_model.rs                 | 1 +
 3 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/crates/agent_ui/src/agent_configuration/add_llm_provider_modal.rs b/crates/agent_ui/src/agent_configuration/add_llm_provider_modal.rs
index b97583377a00d28ea1a8aae6a1380cff3b69e6a0..e0df79ba4dfe226652818b120b7bfcc493c73b1e 100644
--- a/crates/agent_ui/src/agent_configuration/add_llm_provider_modal.rs
+++ b/crates/agent_ui/src/agent_configuration/add_llm_provider_modal.rs
@@ -202,6 +202,7 @@ impl ModelInput {
                 .text(cx)
                 .parse::<u64>()
                 .map_err(|_| SharedString::from("Max Tokens must be a number"))?,
+            reasoning_effort: None,
             capabilities: ModelCapabilities {
                 tools: self.capabilities.supports_tools.selected(),
                 images: self.capabilities.supports_images.selected(),
diff --git a/crates/language_models/src/provider/open_ai_compatible.rs b/crates/language_models/src/provider/open_ai_compatible.rs
index 9f63a1e1a039998c275637f3831b51474c8049ac..1c3268749c3340826cd2f50d29e80eecfa1826d4 100644
--- a/crates/language_models/src/provider/open_ai_compatible.rs
+++ b/crates/language_models/src/provider/open_ai_compatible.rs
@@ -402,7 +402,7 @@ impl LanguageModel for OpenAiCompatibleLanguageModel {
                 self.model.capabilities.parallel_tool_calls,
                 self.model.capabilities.prompt_cache_key,
                 self.max_output_tokens(),
-                None,
+                self.model.reasoning_effort.clone(),
             );
             let completions = self.stream_completion(request, cx);
             async move {
@@ -417,7 +417,7 @@ impl LanguageModel for OpenAiCompatibleLanguageModel {
                 self.model.capabilities.parallel_tool_calls,
                 self.model.capabilities.prompt_cache_key,
                 self.max_output_tokens(),
-                None,
+                self.model.reasoning_effort.clone(),
             );
             let completions = self.stream_response(request, cx);
             async move {
diff --git a/crates/settings_content/src/language_model.rs b/crates/settings_content/src/language_model.rs
index fef92dc8f43d52c160c1e8c8a2fb7aeb0533e2c0..4b72c2ad3f47d834dfa38555d80a8646e3940f51 100644
--- a/crates/settings_content/src/language_model.rs
+++ b/crates/settings_content/src/language_model.rs
@@ -278,6 +278,7 @@ pub struct OpenAiCompatibleAvailableModel {
     pub max_tokens: u64,
     pub max_output_tokens: Option<u64>,
     pub max_completion_tokens: Option<u64>,
+    pub reasoning_effort: Option<OpenAiReasoningEffort>,
     #[serde(default)]
     pub capabilities: OpenAiCompatibleModelCapabilities,
 }