feat(usage): add support for requested model alias handling

- Introduced methods for setting and retrieving model aliases in execution and usage contexts.
- Enhanced `UsageReporter` and related structures to include client-requested aliases.
- Updated tests to validate alias propagation and ensure correct usage reporting.
- Adjusted metadata handling in CLIProxyAPI executors to address alias integration.

internal/redisqueue/plugin.go

@@ -33,6 +33,10 @@ func (p *usageQueuePlugin) HandleUsage(ctx context.Context, record coreusage.Rec
 	if modelName == "" {
 		modelName = "unknown"
 	}
+	aliasName := strings.TrimSpace(record.Alias)
+	if aliasName == "" {
+		aliasName = modelName
+	}
 	provider := strings.TrimSpace(record.Provider)
 	if provider == "" {
 		provider = "unknown"
@@ -76,6 +80,7 @@ func (p *usageQueuePlugin) HandleUsage(ctx context.Context, record coreusage.Rec
 		requestDetail: detail,
 		Provider:      provider,
 		Model:         modelName,
+		Alias:         aliasName,
 		Endpoint:      resolveEndpoint(ctx),
 		AuthType:      authType,
 		APIKey:        apiKey,
@@ -91,6 +96,7 @@ type queuedUsageDetail struct {
 	requestDetail
 	Provider  string `json:"provider"`
 	Model     string `json:"model"`
+	Alias     string `json:"alias"`
 	Endpoint  string `json:"endpoint"`
 	AuthType  string `json:"auth_type"`
 	APIKey    string `json:"api_key"`

internal/redisqueue/plugin_test.go

@@ -24,6 +24,7 @@ func TestUsageQueuePluginPayloadIncludesStableFieldsAndSuccess(t *testing.T) {
 		plugin.HandleUsage(ctx, coreusage.Record{
 			Provider:    "openai",
 			Model:       "gpt-5.4",
+			Alias:       "client-gpt",
 			APIKey:      "test-key",
 			AuthIndex:   "0",
 			AuthType:    "apikey",
@@ -40,6 +41,7 @@ func TestUsageQueuePluginPayloadIncludesStableFieldsAndSuccess(t *testing.T) {
 		payload := popSinglePayload(t)
 		requireStringField(t, payload, "provider", "openai")
 		requireStringField(t, payload, "model", "gpt-5.4")
+		requireStringField(t, payload, "alias", "client-gpt")
 		requireStringField(t, payload, "endpoint", "POST /v1/chat/completions")
 		requireStringField(t, payload, "auth_type", "apikey")
 		requireStringField(t, payload, "request_id", "ctx-request-id")
@@ -58,6 +60,7 @@ func TestUsageQueuePluginPayloadIncludesStableFieldsAndFailureAndGinRequestID(t
 		plugin.HandleUsage(ctx, coreusage.Record{
 			Provider:    "openai",
 			Model:       "gpt-5.4-mini",
+			Alias:       "client-mini",
 			APIKey:      "test-key",
 			AuthIndex:   "0",
 			AuthType:    "apikey",
@@ -74,6 +77,7 @@ func TestUsageQueuePluginPayloadIncludesStableFieldsAndFailureAndGinRequestID(t
 		payload := popSinglePayload(t)
 		requireStringField(t, payload, "provider", "openai")
 		requireStringField(t, payload, "model", "gpt-5.4-mini")
+		requireStringField(t, payload, "alias", "client-mini")
 		requireStringField(t, payload, "endpoint", "GET /v1/responses")
 		requireStringField(t, payload, "auth_type", "apikey")
 		requireStringField(t, payload, "request_id", "gin-request-id")
@@ -102,6 +106,7 @@ func TestUsageQueuePluginAsyncIgnoresRecycledGinContext(t *testing.T) {
 		mgr.Publish(ctx, coreusage.Record{
 			Provider:    "openai",
 			Model:       "gpt-5.4",
+			Alias:       "client-gpt",
 			APIKey:      "test-key",
 			AuthIndex:   "0",
 			AuthType:    "apikey",
@@ -117,6 +122,7 @@ func TestUsageQueuePluginAsyncIgnoresRecycledGinContext(t *testing.T) {
 
 		payload := waitForSinglePayload(t, 2*time.Second)
 		requireStringField(t, payload, "endpoint", "POST /v1/chat/completions")
+		requireStringField(t, payload, "alias", "client-gpt")
 		requireStringField(t, payload, "request_id", "ctx-request-id")
 		requireBoolField(t, payload, "failed", true)
 	})

internal/runtime/executor/helps/usage_helpers.go

@@ -18,6 +18,7 @@ import (
 type UsageReporter struct {
 	provider    string
 	model       string
+	alias       string
 	authID      string
 	authIndex   string
 	authType    string
@@ -29,9 +30,14 @@ type UsageReporter struct {
 
 func NewUsageReporter(ctx context.Context, provider, model string, auth *cliproxyauth.Auth) *UsageReporter {
 	apiKey := APIKeyFromContext(ctx)
+	alias := usage.RequestedModelAliasFromContext(ctx)
+	if alias == "" {
+		alias = model
+	}
 	reporter := &UsageReporter{
 		provider:    provider,
 		model:       model,
+		alias:       strings.TrimSpace(alias),
 		requestedAt: time.Now(),
 		apiKey:      apiKey,
 		source:      resolveUsageSource(auth, apiKey),
@@ -139,6 +145,7 @@ func (r *UsageReporter) buildRecordForModel(model string, detail usage.Detail, f
 	return usage.Record{
 		Provider:    r.provider,
 		Model:       model,
+		Alias:       r.alias,
 		Source:      r.source,
 		APIKey:      r.apiKey,
 		AuthID:      r.authID,

internal/runtime/executor/helps/usage_helpers_test.go

@@ -1,6 +1,7 @@
 package helps
 
 import (
+	"context"
 	"testing"
 	"time"
 
@@ -107,6 +108,19 @@ func TestUsageReporterBuildRecordIncludesLatency(t *testing.T) {
 	}
 }
 
+func TestUsageReporterBuildRecordIncludesRequestedModelAlias(t *testing.T) {
+	ctx := usage.WithRequestedModelAlias(context.Background(), "client-gpt")
+	reporter := NewUsageReporter(ctx, "openai", "gpt-5.4", nil)
+
+	record := reporter.buildRecord(usage.Detail{TotalTokens: 3}, false)
+	if record.Model != "gpt-5.4" {
+		t.Fatalf("model = %q, want %q", record.Model, "gpt-5.4")
+	}
+	if record.Alias != "client-gpt" {
+		t.Fatalf("alias = %q, want %q", record.Alias, "client-gpt")
+	}
+}
+
 func TestUsageReporterBuildAdditionalModelRecordSkipsZeroTokens(t *testing.T) {
 	reporter := &UsageReporter{
 		provider:    "codex",

sdk/api/handlers/handlers.go

@@ -539,7 +539,7 @@ func (h *BaseAPIHandler) ExecuteWithAuthManager(ctx context.Context, handlerType
 		return nil, nil, errMsg
 	}
 	reqMeta := requestExecutionMetadata(ctx)
-	reqMeta[coreexecutor.RequestedModelMetadataKey] = normalizedModel
+	reqMeta[coreexecutor.RequestedModelMetadataKey] = modelName
 	payload := rawJSON
 	if len(payload) == 0 {
 		payload = nil
@@ -587,7 +587,7 @@ func (h *BaseAPIHandler) ExecuteCountWithAuthManager(ctx context.Context, handle
 		return nil, nil, errMsg
 	}
 	reqMeta := requestExecutionMetadata(ctx)
-	reqMeta[coreexecutor.RequestedModelMetadataKey] = normalizedModel
+	reqMeta[coreexecutor.RequestedModelMetadataKey] = modelName
 	payload := rawJSON
 	if len(payload) == 0 {
 		payload = nil
@@ -639,7 +639,7 @@ func (h *BaseAPIHandler) ExecuteStreamWithAuthManager(ctx context.Context, handl
 		return nil, nil, errChan
 	}
 	reqMeta := requestExecutionMetadata(ctx)
-	reqMeta[coreexecutor.RequestedModelMetadataKey] = normalizedModel
+	reqMeta[coreexecutor.RequestedModelMetadataKey] = modelName
 	payload := rawJSON
 	if len(payload) == 0 {
 		payload = nil

sdk/cliproxy/auth/conductor.go

@@ -22,6 +22,7 @@ import (
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/thinking"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 	cliproxyexecutor "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/executor"
+	coreusage "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/usage"
 	log "github.com/sirupsen/logrus"
 )
 
@@ -827,6 +828,7 @@ func (m *Manager) executeStreamWithModelPool(ctx context.Context, executor Provi
 	if executor == nil {
 		return nil, &Error{Code: "executor_not_found", Message: "executor not registered"}
 	}
+	ctx = contextWithRequestedModelAlias(ctx, opts, routeModel)
 	var lastErr error
 	for idx, execModel := range execModels {
 		resultModel := m.stateModelForExecution(auth, routeModel, execModel, pooled)
@@ -1319,6 +1321,7 @@ func (m *Manager) executeMixedOnce(ctx context.Context, providers []string, req
 			execCtx = context.WithValue(execCtx, roundTripperContextKey{}, rt)
 			execCtx = context.WithValue(execCtx, "cliproxy.roundtripper", rt)
 		}
+		execCtx = contextWithRequestedModelAlias(execCtx, opts, routeModel)
 
 		models, pooled := m.preparedExecutionModels(auth, routeModel)
 		if len(models) == 0 {
@@ -1397,6 +1400,7 @@ func (m *Manager) executeCountMixedOnce(ctx context.Context, providers []string,
 			execCtx = context.WithValue(execCtx, roundTripperContextKey{}, rt)
 			execCtx = context.WithValue(execCtx, "cliproxy.roundtripper", rt)
 		}
+		execCtx = contextWithRequestedModelAlias(execCtx, opts, routeModel)
 
 		models, pooled := m.preparedExecutionModels(auth, routeModel)
 		if len(models) == 0 {
@@ -1534,6 +1538,36 @@ func hasRequestedModelMetadata(meta map[string]any) bool {
 	}
 }
 
+func contextWithRequestedModelAlias(ctx context.Context, opts cliproxyexecutor.Options, fallback string) context.Context {
+	alias := requestedModelAliasFromOptions(opts, fallback)
+	return coreusage.WithRequestedModelAlias(ctx, alias)
+}
+
+func requestedModelAliasFromOptions(opts cliproxyexecutor.Options, fallback string) string {
+	fallback = strings.TrimSpace(fallback)
+	if len(opts.Metadata) == 0 {
+		return fallback
+	}
+	raw, ok := opts.Metadata[cliproxyexecutor.RequestedModelMetadataKey]
+	if !ok || raw == nil {
+		return fallback
+	}
+	switch value := raw.(type) {
+	case string:
+		if strings.TrimSpace(value) == "" {
+			return fallback
+		}
+		return strings.TrimSpace(value)
+	case []byte:
+		if len(value) == 0 {
+			return fallback
+		}
+		return strings.TrimSpace(string(value))
+	default:
+		return fallback
+	}
+}
+
 func pinnedAuthIDFromMetadata(meta map[string]any) string {
 	if len(meta) == 0 {
 		return ""
@@ -3096,6 +3130,7 @@ func (m *Manager) tryAntigravityCreditsExecute(ctx context.Context, req cliproxy
 			creditsCtx = context.WithValue(creditsCtx, "cliproxy.roundtripper", rt)
 		}
 		creditsOpts := ensureRequestedModelMetadata(opts, routeModel)
+		creditsCtx = contextWithRequestedModelAlias(creditsCtx, creditsOpts, routeModel)
 		publishSelectedAuthMetadata(creditsOpts.Metadata, c.auth.ID)
 		models := m.executionModelCandidates(c.auth, routeModel)
 		if len(models) == 0 {

sdk/cliproxy/auth/conductor_oauth_alias_suspension_test.go

@@ -10,20 +10,23 @@ import (
 	internalconfig "github.com/router-for-me/CLIProxyAPI/v6/internal/config"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/registry"
 	cliproxyexecutor "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/executor"
+	coreusage "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/usage"
 )
 
 type aliasRoutingExecutor struct {
 	id string
 
-	mu            sync.Mutex
-	executeModels []string
+	mu             sync.Mutex
+	executeModels  []string
+	executeAliases []string
 }
 
 func (e *aliasRoutingExecutor) Identifier() string { return e.id }
 
-func (e *aliasRoutingExecutor) Execute(_ context.Context, _ *Auth, req cliproxyexecutor.Request, _ cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
+func (e *aliasRoutingExecutor) Execute(ctx context.Context, _ *Auth, req cliproxyexecutor.Request, _ cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
 	e.mu.Lock()
 	e.executeModels = append(e.executeModels, req.Model)
+	e.executeAliases = append(e.executeAliases, coreusage.RequestedModelAliasFromContext(ctx))
 	e.mu.Unlock()
 	return cliproxyexecutor.Response{Payload: []byte(req.Model)}, nil
 }
@@ -52,6 +55,14 @@ func (e *aliasRoutingExecutor) ExecuteModels() []string {
 	return out
 }
 
+func (e *aliasRoutingExecutor) ExecuteAliases() []string {
+	e.mu.Lock()
+	defer e.mu.Unlock()
+	out := make([]string, len(e.executeAliases))
+	copy(out, e.executeAliases)
+	return out
+}
+
 func TestManagerExecute_OAuthAliasBypassesBlockedRouteModel(t *testing.T) {
 	const (
 		provider    = "antigravity"
@@ -108,4 +119,12 @@ func TestManagerExecute_OAuthAliasBypassesBlockedRouteModel(t *testing.T) {
 	if gotModels[0] != targetModel {
 		t.Fatalf("execute model = %q, want %q", gotModels[0], targetModel)
 	}
+
+	gotAliases := executor.ExecuteAliases()
+	if len(gotAliases) != 1 {
+		t.Fatalf("execute aliases len = %d, want 1", len(gotAliases))
+	}
+	if gotAliases[0] != routeModel {
+		t.Fatalf("execute alias = %q, want %q", gotAliases[0], routeModel)
+	}
 }

sdk/cliproxy/usage/manager.go

@@ -2,6 +2,7 @@ package usage
 
 import (
 	"context"
+	"strings"
 	"sync"
 	"time"
 
@@ -12,6 +13,7 @@ import (
 type Record struct {
 	Provider    string
 	Model       string
+	Alias       string
 	APIKey      string
 	AuthID      string
 	AuthIndex   string
@@ -32,6 +34,36 @@ type Detail struct {
 	TotalTokens     int64
 }
 
+type requestedModelAliasContextKey struct{}
+
+// WithRequestedModelAlias stores the client-requested model name for usage sinks.
+func WithRequestedModelAlias(ctx context.Context, alias string) context.Context {
+	if ctx == nil {
+		ctx = context.Background()
+	}
+	alias = strings.TrimSpace(alias)
+	if alias == "" {
+		return ctx
+	}
+	return context.WithValue(ctx, requestedModelAliasContextKey{}, alias)
+}
+
+// RequestedModelAliasFromContext returns the client-requested model name stored in ctx.
+func RequestedModelAliasFromContext(ctx context.Context) string {
+	if ctx == nil {
+		return ""
+	}
+	raw := ctx.Value(requestedModelAliasContextKey{})
+	switch value := raw.(type) {
+	case string:
+		return strings.TrimSpace(value)
+	case []byte:
+		return strings.TrimSpace(string(value))
+	default:
+		return ""
+	}
+}
+
 // Plugin consumes usage records emitted by the proxy runtime.
 type Plugin interface {
 	HandleUsage(ctx context.Context, record Record)