Merge branch 'router-for-me:main' into main

internal/runtime/executor/openai_compat_executor.go

@@ -54,13 +54,14 @@ func (e *OpenAICompatExecutor) Execute(ctx context.Context, auth *cliproxyauth.A
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("openai")
 	translated := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), opts.Stream)
-	if modelOverride := e.resolveUpstreamModel(req.Model, auth); modelOverride != "" {
+	modelOverride := e.resolveUpstreamModel(req.Model, auth)
+	if modelOverride != "" {
 		translated = e.overrideModel(translated, modelOverride)
 	}
 	translated = applyPayloadConfigWithRoot(e.cfg, req.Model, to.String(), "", translated)
 	translated = applyReasoningEffortMetadata(translated, req.Metadata, req.Model, "reasoning_effort")
 	upstreamModel := util.ResolveOriginalModel(req.Model, req.Metadata)
-	if upstreamModel != "" {
+	if upstreamModel != "" && modelOverride == "" {
 		translated, _ = sjson.SetBytes(translated, "model", upstreamModel)
 	}
 	translated = normalizeThinkingConfig(translated, upstreamModel)
@@ -148,13 +149,14 @@ func (e *OpenAICompatExecutor) ExecuteStream(ctx context.Context, auth *cliproxy
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("openai")
 	translated := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), true)
-	if modelOverride := e.resolveUpstreamModel(req.Model, auth); modelOverride != "" {
+	modelOverride := e.resolveUpstreamModel(req.Model, auth)
+	if modelOverride != "" {
 		translated = e.overrideModel(translated, modelOverride)
 	}
 	translated = applyPayloadConfigWithRoot(e.cfg, req.Model, to.String(), "", translated)
 	translated = applyReasoningEffortMetadata(translated, req.Metadata, req.Model, "reasoning_effort")
 	upstreamModel := util.ResolveOriginalModel(req.Model, req.Metadata)
-	if upstreamModel != "" {
+	if upstreamModel != "" && modelOverride == "" {
 		translated, _ = sjson.SetBytes(translated, "model", upstreamModel)
 	}
 	translated = normalizeThinkingConfig(translated, upstreamModel)

internal/runtime/executor/payload_helpers.go

@@ -52,10 +52,14 @@ func applyReasoningEffortMetadata(payload []byte, metadata map[string]any, model
 	if len(metadata) == 0 {
 		return payload
 	}
-	if !util.ModelSupportsThinking(model) {
+	if field == "" {
 		return payload
 	}
-	if field == "" {
+	baseModel := util.ResolveOriginalModel(model, metadata)
+	if baseModel == "" {
+		baseModel = model
+	}
+	if !util.ModelSupportsThinking(baseModel) && !util.IsOpenAICompatibilityModel(baseModel) {
 		return payload
 	}
 	if effort, ok := util.ReasoningEffortFromMetadata(metadata); ok && effort != "" {
@@ -226,6 +230,9 @@ func normalizeThinkingConfig(payload []byte, model string) []byte {
 	}
 
 	if !util.ModelSupportsThinking(model) {
+		if util.IsOpenAICompatibilityModel(model) {
+			return payload
+		}
 		return stripThinkingFields(payload)
 	}
 

internal/util/thinking.go

@@ -25,33 +25,33 @@ func ModelSupportsThinking(model string) bool {
 // or min (0 if zero is allowed and mid <= 0).
 func NormalizeThinkingBudget(model string, budget int) int {
 	if budget == -1 { // dynamic
-		if found, min, max, zeroAllowed, dynamicAllowed := thinkingRangeFromRegistry(model); found {
+		if found, minBudget, maxBudget, zeroAllowed, dynamicAllowed := thinkingRangeFromRegistry(model); found {
 			if dynamicAllowed {
 				return -1
 			}
-			mid := (min + max) / 2
+			mid := (minBudget + maxBudget) / 2
 			if mid <= 0 && zeroAllowed {
 				return 0
 			}
 			if mid <= 0 {
-				return min
+				return minBudget
 			}
 			return mid
 		}
 		return -1
 	}
-	if found, min, max, zeroAllowed, _ := thinkingRangeFromRegistry(model); found {
+	if found, minBudget, maxBudget, zeroAllowed, _ := thinkingRangeFromRegistry(model); found {
 		if budget == 0 {
 			if zeroAllowed {
 				return 0
 			}
-			return min
+			return minBudget
 		}
-		if budget < min {
+		if budget < minBudget {
-			return min
+			return minBudget
 		}
-		if budget > max {
+		if budget > maxBudget {
-			return max
+			return maxBudget
 		}
 		return budget
 	}
@@ -105,3 +105,16 @@ func NormalizeReasoningEffortLevel(model, effort string) (string, bool) {
 	}
 	return "", false
 }
+
+// IsOpenAICompatibilityModel reports whether the model is registered as an OpenAI-compatibility model.
+// These models may not advertise Thinking metadata in the registry.
+func IsOpenAICompatibilityModel(model string) bool {
+	if model == "" {
+		return false
+	}
+	info := registry.GetGlobalRegistry().GetModelInfo(model)
+	if info == nil {
+		return false
+	}
+	return strings.EqualFold(strings.TrimSpace(info.Type), "openai-compatibility")
+}