fix(llm): 移除测试用 token 限制，保留完整的推理模型 temperature 兼容与 Usage 空指针修复

internal/llm/client.go

@@ -464,7 +464,7 @@ func (c *Client) ReviewStream(ctx context.Context, systemPrompt, diff string, on
 		}
 
 		// 捕获最后一个 chunk 中携带的 Usage 信息
-		if response.Usage.TotalTokens > 0 {
+		if response.Usage != nil && response.Usage.TotalTokens > 0 {
 			usage.InputTokens = response.Usage.PromptTokens
 			usage.OutputTokens = response.Usage.CompletionTokens
 		}

internal/llm/model_test.go

@@ -0,0 +1,51 @@
+package llm
+
+import (
+	"fmt"
+	"testing"
+)
+
+func TestIsReasoningModel(t *testing.T) {
+	tests := []struct {
+		model    string
+		expected bool
+	}{
+		{"o1", true},
+		{"o1-mini", true},
+		{"o1-preview", true},
+		{"o3-mini", true},
+		{"gpt-5.6-luna", true},
+		{"gpt-5.6-sol", true},
+		{"gpt-5.6-terra", true},
+		{"deepseek-reasoner", true},
+		{"gemini-3.7-flash", false},
+		{"claude-4.6-sonnet", false},
+		{"gpt-4o", false},
+	}
+
+	for _, tt := range tests {
+		got := isReasoningModel(tt.model)
+		if got != tt.expected {
+			t.Errorf("isReasoningModel(%q) = %v, expected %v", tt.model, got, tt.expected)
+		}
+	}
+}
+
+func TestIsTemperatureError(t *testing.T) {
+	tests := []struct {
+		errStr   string
+		expected bool
+	}{
+		{"this model has beta-limitations, temperature, top_p and n are fixed at 1", true},
+		{"temperature must be 1.0", true},
+		{"invalid api key", false},
+		{"rate limit exceeded", false},
+	}
+
+	for _, tt := range tests {
+		got := isTemperatureError(fmt.Errorf("%s", tt.errStr))
+		if got != tt.expected {
+			t.Errorf("isTemperatureError(%q) = %v, expected %v", tt.errStr, got, tt.expected)
+		}
+	}
+}