feat: add StepFun provider support

catalog/live/fetchers.go

@@ -85,6 +85,7 @@ var Registry = map[string]FetchFunc{
 	"poolside":               FetchPoolside,
 	"groq":                   FetchGroq,
 	"clinepass":              FetchClinePass,
+	"stepfun":                FetchStepFun,
 }
 
 // Fetch runs a registered live fetcher.

catalog/live/fetchers_providers.go

@@ -1058,3 +1058,64 @@ func FetchClinePass(env map[string]string) ([]Entry, error) {
 	}
 	return entries, nil
 }
+
+// FetchStepFun lists models from the StepFun OpenAI-compatible API.
+// Docs: https://platform.stepfun.ai/docs/en/guides/basic-concepts
+// Pricing: https://platform.stepfun.ai/docs/en/guides/pricing/details
+func FetchStepFun(env map[string]string) ([]Entry, error) {
+	entries, err := fetchOpenAICompatModels(
+		context.Background(),
+		envOr(env, "STEP_BASE_URL", "https://api.stepfun.ai/v1"),
+		env["STEP_API_KEY"], "Bearer",
+	)
+	if err != nil {
+		return nil, err
+	}
+	for i := range entries {
+		entries[i] = enrichStepFunEntry(entries[i])
+	}
+	return entries, nil
+}
+
+var stepFunOfficialTextSpecs = map[string]struct {
+	ContextWindow, MaxOutput int
+	Tools, Vision, Thinking  bool
+	InputPrice, OutputPrice  float64
+}{
+	"step-3.7-flash":      {256_000, 65_536, true, true, true, 0.20, 1.15},
+	"step-3.5-flash":      {256_000, 65_536, true, false, true, 0.10, 0.30},
+	"step-3.5-flash-2603": {256_000, 65_536, true, false, true, 0.10, 0.30},
+}
+
+func enrichStepFunEntry(e Entry) Entry {
+	id := strings.TrimSpace(e.ID)
+	if i := strings.LastIndex(id, "/"); i >= 0 {
+		id = id[i+1:]
+	}
+	spec, ok := stepFunOfficialTextSpecs[id]
+	if !ok {
+		return e
+	}
+	if e.ContextWindow <= 0 {
+		e.ContextWindow = spec.ContextWindow
+	}
+	if e.MaxOutput <= 0 {
+		e.MaxOutput = spec.MaxOutput
+	}
+	if spec.Tools {
+		e.Features = appendUnique(e.Features, "tools")
+	}
+	if spec.Vision {
+		e.ImageInput = true
+		e.Features = appendUnique(e.Features, "image_input")
+	}
+	if spec.Thinking {
+		e.ThinkingEnabled = true
+		e.Features = appendUnique(e.Features, "thinking:enabled")
+	}
+	if spec.InputPrice >= 0 && spec.OutputPrice >= 0 {
+		e.InputPricePer1M = spec.InputPrice
+		e.OutputPricePer1M = spec.OutputPrice
+	}
+	return e
+}

catalog/provider_live_parity_test.go

@@ -11,8 +11,8 @@ import (
 func TestAllProviders_LiveFetchParity(t *testing.T) {
 	t.Parallel()
 	specs := registry.All()
-	if len(specs) != 25 {
-		t.Fatalf("expected 25 providers, got %d", len(specs))
+	if len(specs) != 26 {
+		t.Fatalf("expected 26 providers, got %d", len(specs))
 	}
 	for _, spec := range specs {
 		t.Run(spec.ProviderID, func(t *testing.T) {

catalog/registry/provider_spec_test.go

@@ -9,8 +9,8 @@ import (
 
 func TestAllProviders_Count(t *testing.T) {
 	t.Parallel()
-	if n := len(registry.All()); n != 25 {
-		t.Fatalf("expected 25 providers, got %d", n)
+	if n := len(registry.All()); n != 26 {
+		t.Fatalf("expected 26 providers, got %d", n)
 	}
 }
 
@@ -42,8 +42,8 @@ func TestProviderSpecs_AgnesOpenAIOnlyLongCatOpenAIPrimary(t *testing.T) {
 func TestLiveFetcherKeys_AllProviders(t *testing.T) {
 	t.Parallel()
 	keys := registry.LiveFetcherKeys()
-	if len(keys) != 25 {
-		t.Fatalf("expected 25 live fetcher keys, got %d", len(keys))
+	if len(keys) != 26 {
+		t.Fatalf("expected 26 live fetcher keys, got %d", len(keys))
 	}
 }
 

catalog/registry/providers.go

@@ -173,6 +173,14 @@ func providerSpecs() []ProviderSpec {
 			PublicModelCatalog: true,
 			ProtocolID:         "openai-responses", AdapterID: "concentrate-responses", RuntimeProfileKey: "concentrate",
 		},
+		{
+			ProviderID: "stepfun", DisplayName: "StepFun", DeploymentID: "stepfun-direct", SortOrder: 2, ChatPreference: 27,
+			RequiresKey: true, CredentialEnv: "STEP_API_KEY",
+			BaseURLEnv: []string{"STEP_BASE_URL"},
+			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "https://api.stepfun.ai/v1",
+			LiveFetcherKey: "stepfun", LiveCatalogKey: "stepfun",
+			ProtocolID: "openai-chat-completions", AdapterID: "openai", RuntimeProfileKey: "stepfun",
+		},
 		{
 			ProviderID: "agnes", DisplayName: "Agnes AI", DeploymentID: "agnes-direct", SortOrder: 1, ChatPreference: 25,
 			RequiresKey: true, CredentialEnv: "AGNES_API_KEY",

catalog/v1_defaults.go

@@ -26,6 +26,7 @@
 		"xiaomi_mimo_payg":       {ID: "xiaomi_mimo_payg", Name: "Xiaomi MiMo (Pay-as-you-go)"},
 		"xiaomi_mimo_token_plan": {ID: "xiaomi_mimo_token_plan", Name: "Xiaomi MiMo (Token Plan)"},
 		"deepseek":               {ID: "deepseek", Name: "DeepSeek"},
+		"stepfun":                {ID: "stepfun", Name: "StepFun"},
 	}
 }
 
@@ -59,6 +60,7 @@
 		"xiaomi_mimo_payg-direct":       deployment("xiaomi_mimo_payg-direct", "Xiaomi MiMo Pay-as-you-go", "xiaomi_mimo_payg", "openai-chat-completions", "xiaomi_mimo", NativeModelIDDiscovered),
 		"xiaomi_mimo_token_plan-direct": deployment("xiaomi_mimo_token_plan-direct", "Xiaomi MiMo Token Plan", "xiaomi_mimo_token_plan", "openai-chat-completions", "xiaomi_mimo", NativeModelIDDiscovered),
 		"deepseek-direct":               deployment("deepseek-direct", "DeepSeek", "deepseek", "openai-chat-completions", "deepseek", NativeModelIDCatalogKnown),
+		"stepfun-direct":                deployment("stepfun-direct", "StepFun", "stepfun", "openai-chat-completions", "openai", NativeModelIDDiscovered),
 	}
 }
 
@@ -122,7 +124,7 @@
 }
 
 // DefaultOfferingTemplates returns offering templates for Azure deployments (model mappings required).
 func DefaultOfferingTemplates(generatedAt time.Time) []ModelOfferingTemplate {
 	var out []ModelOfferingTemplate
 	for _, model := range seedOpenAIModels {
 		modelID := "openai/" + model.ID

config/config_test.go

@@ -87,8 +87,8 @@ func TestNormalizeOllamaOpenAIBaseURL(t *testing.T) {
 
 func TestProviderDetectionOrder(t *testing.T) {
 	t.Parallel()
-	if len(APIProviderDetectionOrder) != 23 {
-		t.Errorf("expected 23 providers in detection order, got %d", len(APIProviderDetectionOrder))
+	if len(APIProviderDetectionOrder) != 24 {
+		t.Errorf("expected 24 providers in detection order, got %d", len(APIProviderDetectionOrder))
 	}
 	if APIProviderDetectionOrder[0] != ProviderAnthropic {
 		t.Error("expected anthropic first in detection order")

config/profiles.go

@@ -30,6 +30,7 @@ const (
 	ProviderPoolside            APIProvider = "poolside"
 	ProviderGroq                APIProvider = "groq"
 	ProviderClinePass           APIProvider = "clinepass"
+	ProviderStepFun             APIProvider = "stepfun"
 )
 
 // RuntimeProviderProfile defines how a provider is detected and configured at runtime.
@@ -212,13 +213,20 @@ var (
 		BaseURLEnv:   []string{"CLINE_API_BASE"},
 		APIKeys:      []APIKeyDef{{Env: "CLINE_API_KEY", Source: "clinepass"}},
 	}
+	StepFunRuntimeProfile = RuntimeProviderProfile{
+		Mode: "openai", DefaultBaseURL: DefaultStepFunOpenAIBaseURL,
+		DetectionEnv: []string{"STEP_API_KEY"},
+		ModelEnv:     []string{"STEP_MODEL", "OPENAI_MODEL"},
+		BaseURLEnv:   []string{"STEP_BASE_URL"},
+		APIKeys:      []APIKeyDef{{Env: "STEP_API_KEY", Source: "stepfun"}},
+	}
 )
 
 // APIProviderDetectionOrder is the priority order for provider detection.
 var APIProviderDetectionOrder = []APIProvider{
 	ProviderAnthropic, ProviderConcentrate, ProviderOpenRouter, ProviderGrok, ProviderGemini,
 	ProviderVertex, ProviderBedrock, ProviderZAICoding, ProviderZAIPayg, ProviderCanopyWave, ProviderDeepSeek, ProviderPoolside, ProviderGroq, ProviderClinePass, ProviderAzure, ProviderOpenAI, ProviderOpenCodeGo,
-	ProviderKimi, ProviderXiaomiMimoPayg, ProviderXiaomiMimoTokenPlan, ProviderMiniMaxTokenPlan, ProviderMiniMaxPayg, ProviderOllama,
+	ProviderKimi, ProviderXiaomiMimoPayg, ProviderXiaomiMimoTokenPlan, ProviderMiniMaxTokenPlan, ProviderMiniMaxPayg, ProviderOllama, ProviderStepFun,
 }
 
 // ProviderModelEnvKeys maps each provider to its model env var keys.
@@ -246,6 +254,7 @@ var ProviderModelEnvKeys = map[APIProvider][]string{
 	ProviderXiaomiMimoTokenPlan: XiaomiTokenPlanRuntimeProfile.ModelEnv,
 	ProviderMiniMaxTokenPlan:    {"MINIMAX_TOKEN_PLAN_MODEL", "MINIMAX_MODEL", "OPENAI_MODEL"},
 	ProviderMiniMaxPayg:         {"MINIMAX_PAYG_MODEL", "MINIMAX_MODEL", "OPENAI_MODEL"},
+	ProviderStepFun:             StepFunRuntimeProfile.ModelEnv,
 }
 
 const (
@@ -255,7 +264,7 @@ const (
 
 // OpenAICompatibleRuntimeProfileOrder is the detection order for runtime profiles.
 var OpenAICompatibleRuntimeProfileOrder = []string{
-	"concentrate", "agnes", "longcat", "openrouter", "grok", "gemini", "anthropic", "zai_coding", "zai_payg", "canopywave", "deepseek", "poolside", "groq", "clinepass", "openai", "opencodego", "kimi", "xiaomi_mimo_payg", "xiaomi_mimo_token_plan", "minimax_token_plan", "minimax_payg",
+	"concentrate", "agnes", "longcat", "openrouter", "grok", "gemini", "anthropic", "zai_coding", "zai_payg", "canopywave", "deepseek", "poolside", "groq", "clinepass", "openai", "opencodego", "kimi", "xiaomi_mimo_payg", "xiaomi_mimo_token_plan", "minimax_token_plan", "minimax_payg", "stepfun",
 }
 
 // OpenAICompatibleRuntimeProfiles maps profile key to its runtime profile.
@@ -281,6 +290,7 @@ var OpenAICompatibleRuntimeProfiles = map[string]RuntimeProviderProfile{
 	"xiaomi_mimo_token_plan": XiaomiTokenPlanRuntimeProfile,
 	"minimax_token_plan":     MiniMaxTokenPlanRuntimeProfile,
 	"minimax_payg":           MiniMaxPaygRuntimeProfile,
+	"stepfun":                StepFunRuntimeProfile,
 }
 
 // RuntimeProviderProfiles maps provider/profile keys to runtime detection profiles.
@@ -308,6 +318,7 @@ var RuntimeProviderProfiles = map[string]RuntimeProviderProfile{
 	"xiaomi_mimo_token_plan": XiaomiTokenPlanRuntimeProfile,
 	"minimax_token_plan":     MiniMaxTokenPlanRuntimeProfile,
 	"minimax_payg":           MiniMaxPaygRuntimeProfile,
+	"stepfun":                StepFunRuntimeProfile,
 	"ollama":                 OllamaRuntimeProfile,
 }
 

config/provider_env.go

@@ -63,6 +63,9 @@ type ProviderConfig struct {
 	ClinePassAPIKey            string                      `json:"clinepass_api_key,omitempty"`
 	ClinePassBaseURL           string                      `json:"clinepass_base_url,omitempty"`
 	ClinePassModel             string                      `json:"clinepass_model,omitempty"`
+	StepFunAPIKey              string                      `json:"stepfun_api_key,omitempty"`
+	StepFunBaseURL             string                      `json:"stepfun_base_url,omitempty"`
+	StepFunModel               string                      `json:"stepfun_model,omitempty"`
 	MiniMaxModel               string                      `json:"minimax_model,omitempty"`
 	AnthropicModel             string                      `json:"anthropic_model,omitempty"`
 	OpenAIModel                string                      `json:"openai_model,omitempty"`
@@ -241,6 +244,11 @@ var providerFields = map[string]providerFieldMap{
 		Models:  func(c *ProviderConfig) []string { return []string{c.XiaomiModel} },
 		BaseURL: func(c *ProviderConfig) string { return c.XiaomiMimoTokenPlanBaseURL },
 	},
+	ProviderStepFun: {
+		APIKeys: func(c *ProviderConfig) []string { return []string{c.StepFunAPIKey} },
+		Models:  func(c *ProviderConfig) []string { return []string{c.StepFunModel} },
+		BaseURL: func(c *ProviderConfig) string { return c.StepFunBaseURL },
+	},
 }
 
 func firstNonEmpty(ss ...string) string {
@@ -587,6 +595,7 @@ func ClearProviderRuntimeEnv() {
 		"POOLSIDE_API_KEY", "POOLSIDE_MODEL", "POOLSIDE_BASE_URL",
 		"CLINE_API_KEY", "CLINE_MODEL", "CLINE_API_BASE",
 		"MOONSHOT_API_KEY", "MOONSHOT_MODEL", "MOONSHOT_BASE_URL",
+		"STEP_API_KEY", "STEP_MODEL", "STEP_BASE_URL",
 		"XIAOMI_MIMO_PAYG_API_KEY", "XIAOMI_MIMO_TOKEN_PLAN_API_KEY",
 		"XIAOMI_MIMO_TOKEN_PLAN_REGION", "XIAOMI_MODEL", "XIAOMI_BASE_URL",
 		"XIAOMI_MIMO_PAYG_BASE_URL", "XIAOMI_MIMO_TOKEN_PLAN_BASE_URL",
@@ -823,6 +832,14 @@ func ApplyProviderEnv(provider string, config *ProviderConfig, activeModel strin
 		if base != "" {
 			collectOpenAICompatibleProvider(env, "XIAOMI", apiKey, m, base, overwrite)
 		}
+	case ProviderStepFun:
+		apiKey := AsNonEmptyString(config.StepFunAPIKey)
+		base := firstNonEmpty(config.StepFunBaseURL, DefaultStepFunOpenAIBaseURL)
+		m := activeModel
+		if m == "" {
+			m = catalog.GetProviderDefaultModel("stepfun", cat)
+		}
+		collectOpenAICompatibleProvider(env, "STEP", apiKey, m, base, overwrite)
 	}
 	return env
 }

config/providers.go

@@ -37,6 +37,7 @@ const (
 	DefaultGroqOpenAIBaseURL        = "https://api.groq.com/openai/v1"
 	DefaultPoolsideOpenAIBaseURL    = "https://inference.poolside.ai/v1"
 	DefaultClinePassOpenAIBaseURL   = "https://api.cline.bot/api/v1" // #nosec G101 -- public API base URL, not a secret value
+	DefaultStepFunOpenAIBaseURL     = "https://api.stepfun.ai/v1"
 )
 
 // ProviderTransport is the transport type for provider requests.