From 1d72a6afa67650d496e0dde4e229ba9a92de893e Mon Sep 17 00:00:00 2001
From: BeaconCat <BeaconCat@users.noreply.github.com>
Date: Tue, 14 Apr 2026 20:29:39 +0800
Subject: [PATCH] feat: --no-thinking adapts to llama.cpp, Ollama, and GLM
 backends

Send all three disable-thinking fields simultaneously:
- chat_template_kwargs.enable_thinking=false (llama.cpp, GLM)
- think=false (Ollama 0.9+)
- thinking.type=disabled (GLM/Zhipu)
Each backend picks the field it recognizes and ignores the rest.
Also bumps max_tokens from 512 to 2048 for thinking models.
---
 cmd/membench/llm_client.go | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/cmd/membench/llm_client.go b/cmd/membench/llm_client.go
index f84173192..2cccd95b7 100644
--- a/cmd/membench/llm_client.go
+++ b/cmd/membench/llm_client.go
@@ -58,7 +58,9 @@ type chatRequest struct {
 	Messages           []chatMessage  `json:"messages"`
 	Temperature        float64        `json:"temperature"`
 	MaxTokens          int            `json:"max_tokens"`
-	ChatTemplateKwargs map[string]any `json:"chat_template_kwargs,omitempty"`
+	ChatTemplateKwargs map[string]any `json:"chat_template_kwargs,omitempty"` // llama.cpp
+	Think              *bool          `json:"think,omitempty"`                // Ollama
+	Thinking           map[string]any `json:"thinking,omitempty"`             // GLM (智谱)
 }
 
 type chatMessage struct {
@@ -86,12 +88,20 @@ func (c *LLMClient) Complete(ctx context.Context, systemPrompt, userPrompt strin
 		Model:       c.Model,
 		Messages:    messages,
 		Temperature: 0.1,
-		MaxTokens:   512,
+		MaxTokens:   2048,
 	}
 	if c.NoThinking {
+		// llama.cpp: chat_template_kwargs
 		body.ChatTemplateKwargs = map[string]any{
 			"enable_thinking": false,
 		}
+		// Ollama (0.9+): think field
+		thinkFalse := false
+		body.Think = &thinkFalse
+		// GLM (智谱): thinking field
+		body.Thinking = map[string]any{
+			"type": "disabled",
+		}
 	}
 
 	jsonBody, err := json.Marshal(body)