feat(models): add Thinking support to GitHub Copilot models

Enhance the model definitions by introducing Thinking support with various levels for each model.

internal/registry/model_definitions.go

@@ -144,6 +144,7 @@ func GetGitHubCopilotModels() []*ModelInfo {
 			ContextLength:       200000,
 			MaxCompletionTokens: 32768,
 			SupportedEndpoints:  []string{"/chat/completions", "/responses"},
+			Thinking:            &ThinkingSupport{Levels: []string{"low", "medium", "high"}},
 		},
 		{
 			ID:                  "gpt-5-mini",
@@ -156,6 +157,7 @@ func GetGitHubCopilotModels() []*ModelInfo {
 			ContextLength:       128000,
 			MaxCompletionTokens: 16384,
 			SupportedEndpoints:  []string{"/chat/completions", "/responses"},
+			Thinking:            &ThinkingSupport{Levels: []string{"low", "medium", "high"}},
 		},
 		{
 			ID:                  "gpt-5-codex",
@@ -168,6 +170,7 @@ func GetGitHubCopilotModels() []*ModelInfo {
 			ContextLength:       200000,
 			MaxCompletionTokens: 32768,
 			SupportedEndpoints:  []string{"/responses"},
+			Thinking:            &ThinkingSupport{Levels: []string{"low", "medium", "high"}},
 		},
 		{
 			ID:                  "gpt-5.1",
@@ -180,6 +183,7 @@ func GetGitHubCopilotModels() []*ModelInfo {
 			ContextLength:       200000,
 			MaxCompletionTokens: 32768,
 			SupportedEndpoints:  []string{"/chat/completions", "/responses"},
+			Thinking:            &ThinkingSupport{Levels: []string{"none", "low", "medium", "high", "xhigh"}},
 		},
 		{
 			ID:                  "gpt-5.1-codex",
@@ -192,6 +196,7 @@ func GetGitHubCopilotModels() []*ModelInfo {
 			ContextLength:       200000,
 			MaxCompletionTokens: 32768,
 			SupportedEndpoints:  []string{"/responses"},
+			Thinking:            &ThinkingSupport{Levels: []string{"none", "low", "medium", "high", "xhigh"}},
 		},
 		{
 			ID:                  "gpt-5.1-codex-mini",
@@ -204,6 +209,7 @@ func GetGitHubCopilotModels() []*ModelInfo {
 			ContextLength:       128000,
 			MaxCompletionTokens: 16384,
 			SupportedEndpoints:  []string{"/responses"},
+			Thinking:            &ThinkingSupport{Levels: []string{"none", "low", "medium", "high", "xhigh"}},
 		},
 		{
 			ID:                  "gpt-5.1-codex-max",
@@ -216,6 +222,7 @@ func GetGitHubCopilotModels() []*ModelInfo {
 			ContextLength:       200000,
 			MaxCompletionTokens: 32768,
 			SupportedEndpoints:  []string{"/responses"},
+			Thinking:            &ThinkingSupport{Levels: []string{"none", "low", "medium", "high", "xhigh"}},
 		},
 		{
 			ID:                  "gpt-5.2",
@@ -228,6 +235,7 @@ func GetGitHubCopilotModels() []*ModelInfo {
 			ContextLength:       200000,
 			MaxCompletionTokens: 32768,
 			SupportedEndpoints:  []string{"/chat/completions", "/responses"},
+			Thinking:            &ThinkingSupport{Levels: []string{"none", "low", "medium", "high", "xhigh"}},
 		},
 		{
 			ID:                  "gpt-5.2-codex",
@@ -240,6 +248,7 @@ func GetGitHubCopilotModels() []*ModelInfo {
 			ContextLength:       200000,
 			MaxCompletionTokens: 32768,
 			SupportedEndpoints:  []string{"/responses"},
+			Thinking:            &ThinkingSupport{Levels: []string{"none", "low", "medium", "high", "xhigh"}},
 		},
 		{
 			ID:                  "claude-haiku-4.5",

internal/runtime/executor/github_copilot_executor.go

@@ -123,6 +123,16 @@ func (e *GitHubCopilotExecutor) Execute(ctx context.Context, auth *cliproxyauth.
 	body := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), false)
 	body = e.normalizeModel(req.Model, body)
 	body = flattenAssistantContent(body)
+
+	thinkingProvider := "openai"
+	if useResponses {
+		thinkingProvider = "codex"
+	}
+	body, err = thinking.ApplyThinking(body, req.Model, from.String(), thinkingProvider, e.Identifier())
+	if err != nil {
+		return resp, err
+	}
+
 	requestedModel := payloadRequestedModel(opts, req.Model)
 	body = applyPayloadConfigWithRoot(e.cfg, req.Model, to.String(), "", body, originalTranslated, requestedModel)
 	body, _ = sjson.SetBytes(body, "stream", false)
@@ -229,6 +239,16 @@ func (e *GitHubCopilotExecutor) ExecuteStream(ctx context.Context, auth *cliprox
 	body := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), true)
 	body = e.normalizeModel(req.Model, body)
 	body = flattenAssistantContent(body)
+
+	thinkingProvider := "openai"
+	if useResponses {
+		thinkingProvider = "codex"
+	}
+	body, err = thinking.ApplyThinking(body, req.Model, from.String(), thinkingProvider, e.Identifier())
+	if err != nil {
+		return nil, err
+	}
+
 	requestedModel := payloadRequestedModel(opts, req.Model)
 	body = applyPayloadConfigWithRoot(e.cfg, req.Model, to.String(), "", body, originalTranslated, requestedModel)
 	body, _ = sjson.SetBytes(body, "stream", true)