chore: merge origin/main into submodule (StepFun + local commits)

AGENTS.md

@@ -8,10 +8,7 @@ When starting any new work (feature, fix, refactor, chore), always create a feat
 
 ## Design Principles
 
-- **Model-agnostic** — single interface for 23 registered LLM providers
-- **Plan-accurate provider metadata** — Concentrate AI is pay-as-you-go, uses
-  `concentrate-payg` as its deployment identifier, and uses its native Responses
-  API (`/v1/responses`) without legacy Chat Completions or Messages routing
+- **Model-agnostic** — single interface for 75+ LLM providers
 - **Host-neutral engine** — Eyrie owns provider routing, transport, caching,
   retry/fallback, and normalized telemetry; hosts own product UX and semantics
 - **Streaming-first** — all responses are streamed; blocking is opt-in

catalog/live_enrich.go

@@ -56,8 +56,8 @@ func FetchLiveProviderCatalog(env map[string]string) (Catalog, []LiveProviderEnr
 				ID:                    deploymentID,
 				Name:                  providerID,
 				ProviderID:            providerID,
-				APIProtocolID:         spec.ProtocolID,
-				AdapterConstructor:    spec.AdapterID,
+				APIProtocolID:         "openai-chat-completions",
+				AdapterConstructor:    "openai",
 				NativeModelIDSource:   NativeModelIDDiscovered,
 				ModelMappingsRequired: false,
 			}
@@ -83,7 +83,7 @@ func FetchLiveProviderCatalog(env map[string]string) (Catalog, []LiveProviderEnr
 				} else if hasInputPricing(entry.RawJSON) {
 					canonicalID = providerID + "/" + entryID
 				}
-			} else {
+			} else if hasInputPricing(entry.RawJSON) {
 				canonicalID = providerID + "/" + entryID
 			}
 
@@ -123,7 +123,7 @@ func FetchLiveModelEntriesForProvider(env map[string]string, providerID string)
 		return nil, fmt.Errorf("catalog: provider %q has no live model list API", providerID)
 	}
 	env = registry.ScopedProviderEnv(spec, env)
-	if !spec.PublicModelCatalog && !registry.CredentialPresent(spec, env) {
+	if !registry.CredentialPresent(spec, env) {
 		return nil, fmt.Errorf("catalog: set %s for %s", spec.CredentialEnv, providerID)
 	}
 	entries, err := live.Fetch(spec.LiveFetcherKey, env)

catalog/registry/providers.go

@@ -17,7 +17,7 @@ func providerSpecs() []ProviderSpec {
 	return []ProviderSpec{
 		// ── Direct API providers ──────────────────────────────────────────
 		{
-			ProviderID: "anthropic", DisplayName: "Anthropic", DeploymentID: "anthropic-direct", SortOrder: 3, ChatPreference: 2,
+			ProviderID: "anthropic", DisplayName: "Anthropic", DeploymentID: "anthropic-direct", SortOrder: 1, ChatPreference: 2,
 			TransportKind: "anthropic",
 			RequiresKey:   true, CredentialEnv: "ANTHROPIC_API_KEY",
 			CredentialAliases: []string{"CLAUDE_API_KEY"},
@@ -28,7 +28,7 @@ func providerSpecs() []ProviderSpec {
 			DirectFallbacks: []string{"openai"},
 		},
 		{
-			ProviderID: "openai", DisplayName: "OpenAI", DeploymentID: "openai-direct", SortOrder: 15, ChatPreference: 1,
+			ProviderID: "openai", DisplayName: "OpenAI", DeploymentID: "openai-direct", SortOrder: 2, ChatPreference: 1,
 			TransportKind: "openai",
 			RequiresKey:   true, CredentialEnv: "OPENAI_API_KEY",
 			BaseURLEnv: []string{"OPENAI_BASE_URL", "OPENAI_API_BASE"},
@@ -38,7 +38,7 @@ func providerSpecs() []ProviderSpec {
 			DirectFallbacks: []string{"anthropic"},
 		},
 		{
-			ProviderID: "gemini", DisplayName: "Gemini API", DeploymentID: "gemini-direct", SortOrder: 9, ChatPreference: 5,
+			ProviderID: "gemini", DisplayName: "Gemini API", DeploymentID: "gemini-direct", SortOrder: 3, ChatPreference: 5,
 			RuntimeBaseURL: "https://generativelanguage.googleapis.com/v1beta/openai",
 			RequiresKey:    true, CredentialEnv: "GEMINI_API_KEY",
 			CredentialAliases: []string{"GOOGLE_API_KEY"},
@@ -48,31 +48,31 @@ func providerSpecs() []ProviderSpec {
 			ProtocolID: "gemini-generate-content", AdapterID: "gemini", RuntimeProfileKey: "gemini",
 		},
 		{
-			ProviderID: "deepseek", DisplayName: "DeepSeek", DeploymentID: "deepseek-direct", SortOrder: 8, ChatPreference: 11,
+			ProviderID: "deepseek", DisplayName: "DeepSeek", DeploymentID: "deepseek-direct", SortOrder: 4, ChatPreference: 11,
 			RequiresKey: true, CredentialEnv: "DEEPSEEK_API_KEY",
 			BaseURLEnv: []string{"DEEPSEEK_BASE_URL"},
 			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "https://api.deepseek.com/v1",
 			LiveFetcherKey: "deepseek", LiveCatalogKey: "deepseek",
 			ProtocolID: "openai-chat-completions", AdapterID: "deepseek", RuntimeProfileKey: "deepseek",
 		},
 		{
-			ProviderID: "grok", DisplayName: "xAI", DeploymentID: "grok-direct", SortOrder: 21, ChatPreference: 4,
+			ProviderID: "grok", DisplayName: "xAI", DeploymentID: "grok-direct", SortOrder: 5, ChatPreference: 4,
 			RequiresKey: true, CredentialEnv: "XAI_API_KEY",
 			BaseURLEnv: []string{"XAI_BASE_URL"},
 			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "https://api.x.ai/v1",
 			LiveFetcherKey: "grok", LiveCatalogKey: "grok",
 			ProtocolID: "openai-chat-completions", AdapterID: "grok", RuntimeProfileKey: "grok",
 		},
 		{
-			ProviderID: "kimi", DisplayName: "Kimi", DeploymentID: "kimi-direct", SortOrder: 11, ChatPreference: 14,
+			ProviderID: "kimi", DisplayName: "Kimi", DeploymentID: "kimi-direct", SortOrder: 6, ChatPreference: 14,
 			RequiresKey: true, CredentialEnv: "MOONSHOT_API_KEY",
 			BaseURLEnv: []string{"MOONSHOT_BASE_URL"},
 			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "https://api.moonshot.ai/v1",
 			LiveFetcherKey: "kimi", LiveCatalogKey: "kimi",
 			ProtocolID: "openai-chat-completions", AdapterID: "kimi", RuntimeProfileKey: "kimi",
 		},
 		{
-			ProviderID: "zai_coding", DisplayName: "Z.AI — Coding Plan", DeploymentID: "zai_coding-direct", SortOrder: 24, ChatPreference: 8,
+			ProviderID: "zai_coding", DisplayName: "Z.AI — Coding Plan", DeploymentID: "zai_coding-direct", SortOrder: 7, ChatPreference: 8,
 			RequiresKey: true, CredentialEnv: "ZAI_CODING_API_KEY",
 			BaseURLEnv: []string{"ZAI_CODING_BASE_URL", "ZAI_BASE_URL"},
 			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "https://api.z.ai/api/coding/paas/v4",
@@ -81,7 +81,7 @@ func providerSpecs() []ProviderSpec {
 			PrepareCredentialEnv: true,
 		},
 		{
-			ProviderID: "zai_payg", DisplayName: "Z.AI — Pay-as-you-go", DeploymentID: "zai_payg-direct", SortOrder: 25, ChatPreference: 9,
+			ProviderID: "zai_payg", DisplayName: "Z.AI — Pay-as-you-go", DeploymentID: "zai_payg-direct", SortOrder: 8, ChatPreference: 9,
 			RequiresKey: true, CredentialEnv: "ZAI_API_KEY",
 			BaseURLEnv: []string{"ZAI_BASE_URL", "ZAI_API_BASE"},
 			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "https://api.z.ai/api/paas/v4",
@@ -90,7 +90,7 @@ func providerSpecs() []ProviderSpec {
 			PrepareCredentialEnv: true,
 		},
 		{
-			ProviderID: "xiaomi_mimo_token_plan", DisplayName: "Xiaomi MiMo — Token Plan", DeploymentID: "xiaomi_mimo_token_plan-direct", SortOrder: 23, ChatPreference: 16,
+			ProviderID: "xiaomi_mimo_token_plan", DisplayName: "Xiaomi MiMo — Token Plan", DeploymentID: "xiaomi_mimo_token_plan-direct", SortOrder: 9, ChatPreference: 16,
 			RequiresKey: true, CredentialEnv: "XIAOMI_MIMO_TOKEN_PLAN_API_KEY",
 			BaseURLEnv: []string{"XIAOMI_MIMO_TOKEN_PLAN_BASE_URL"},
 			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "",
@@ -99,7 +99,7 @@ func providerSpecs() []ProviderSpec {
 			PrepareCredentialEnv: true,
 		},
 		{
-			ProviderID: "xiaomi_mimo_payg", DisplayName: "Xiaomi MiMo — Pay-as-you-go", DeploymentID: "xiaomi_mimo_payg-direct", SortOrder: 22, ChatPreference: 15,
+			ProviderID: "xiaomi_mimo_payg", DisplayName: "Xiaomi MiMo — Pay-as-you-go", DeploymentID: "xiaomi_mimo_payg-direct", SortOrder: 10, ChatPreference: 15,
 			RequiresKey: true, CredentialEnv: "XIAOMI_MIMO_PAYG_API_KEY",
 			CredentialAliases: []string{"XIAOMI_MIMO_API_KEY"},
 			BaseURLEnv:        []string{"XIAOMI_MIMO_PAYG_BASE_URL", "XIAOMI_BASE_URL"},
@@ -108,15 +108,15 @@ func providerSpecs() []ProviderSpec {
 			ProtocolID: "openai-chat-completions", AdapterID: "xiaomi_mimo", RuntimeProfileKey: "xiaomi_mimo_payg",
 		},
 		{
-			ProviderID: "minimax_token_plan", DisplayName: "MiniMax — Token Plan", DeploymentID: "minimax_token_plan-direct", SortOrder: 14, ChatPreference: 17,
+			ProviderID: "minimax_token_plan", DisplayName: "MiniMax — Token Plan", DeploymentID: "minimax_token_plan-direct", SortOrder: 11, ChatPreference: 17,
 			RequiresKey: true, CredentialEnv: "MINIMAX_TOKEN_PLAN_API_KEY",
 			BaseURLEnv: []string{"MINIMAX_TOKEN_PLAN_BASE_URL"},
 			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "https://api.minimax.io/v1",
 			LiveFetcherKey: "minimax_token_plan", LiveCatalogKey: "minimax_token_plan",
 			ProtocolID: "openai-chat-completions", AdapterID: "openai", RuntimeProfileKey: "minimax_token_plan",
 		},
 		{
-			ProviderID: "minimax_payg", DisplayName: "MiniMax — Pay-as-you-go", DeploymentID: "minimax_payg-direct", SortOrder: 13, ChatPreference: 18,
+			ProviderID: "minimax_payg", DisplayName: "MiniMax — Pay-as-you-go", DeploymentID: "minimax_payg-direct", SortOrder: 12, ChatPreference: 18,
 			RequiresKey: true, CredentialEnv: "MINIMAX_PAYG_API_KEY",
 			BaseURLEnv: []string{"MINIMAX_PAYG_BASE_URL", "MINIMAX_BASE_URL"},
 			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "https://api.minimax.io/v1",
@@ -126,7 +126,7 @@ func providerSpecs() []ProviderSpec {
 
 		// ── Cloud platform providers ──────────────────────────────────────
 		{
-			ProviderID: "azure", DisplayName: "Azure OpenAI", DeploymentID: "openai-azure", SortOrder: 4, ChatPreference: 12,
+			ProviderID: "azure", DisplayName: "Azure OpenAI", DeploymentID: "openai-azure", SortOrder: 13, ChatPreference: 12,
 			TransportKind: "azure",
 			RequiresKey:   true, CredentialEnv: "AZURE_OPENAI_API_KEY",
 			BaseURLEnv:     []string{"AZURE_OPENAI_ENDPOINT"},
@@ -135,7 +135,7 @@ func providerSpecs() []ProviderSpec {
 			ProtocolID: "openai-chat-completions", AdapterID: "openai-azure", RuntimeProfileKey: "azure",
 		},
 		{
-			ProviderID: "bedrock", DisplayName: "Amazon Bedrock", DeploymentID: "anthropic-bedrock", SortOrder: 2, ChatPreference: 7,
+			ProviderID: "bedrock", DisplayName: "Amazon Bedrock", DeploymentID: "anthropic-bedrock", SortOrder: 14, ChatPreference: 7,
 			TransportKind: "bedrock",
 			RequiresKey:   true, CredentialEnv: "AWS_SECRET_ACCESS_KEY",
 			CredentialEnvFallbacks: []string{"AWS_ACCESS_KEY_ID", "AWS_SESSION_TOKEN"},
@@ -145,7 +145,7 @@ func providerSpecs() []ProviderSpec {
 			ProtocolID: "anthropic-messages", AdapterID: "anthropic-bedrock", RuntimeProfileKey: "bedrock",
 		},
 		{
-			ProviderID: "vertex", DisplayName: "Vertex AI", DeploymentID: "gemini-vertex", SortOrder: 20, ChatPreference: 6,
+			ProviderID: "vertex", DisplayName: "Vertex AI", DeploymentID: "gemini-vertex", SortOrder: 15, ChatPreference: 6,
 			TransportKind: "vertex",
 			RequiresKey:   true, CredentialEnv: "VERTEX_ACCESS_TOKEN",
 			CredentialEnvFallbacks: []string{"GOOGLE_OAUTH_ACCESS_TOKEN"},
@@ -157,7 +157,7 @@ func providerSpecs() []ProviderSpec {
 
 		// ── Aggregators ───────────────────────────────────────────────────
 		{
-			ProviderID: "openrouter", DisplayName: "OpenRouter", DeploymentID: "openrouter", SortOrder: 17, ChatPreference: 3,
+			ProviderID: "openrouter", DisplayName: "OpenRouter", DeploymentID: "openrouter", SortOrder: 16, ChatPreference: 3,
 			RequiresKey: true, CredentialEnv: "OPENROUTER_API_KEY",
 			BaseURLEnv: []string{"OPENROUTER_BASE_URL"},
 			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "https://openrouter.ai/api/v1",
@@ -200,31 +200,31 @@ func providerSpecs() []ProviderSpec {
 
 		// ── Niche ─────────────────────────────────────────────────────────
 		{
-			ProviderID: "canopywave", DisplayName: "CanopyWave", DeploymentID: "canopywave", SortOrder: 5, ChatPreference: 10,
+			ProviderID: "canopywave", DisplayName: "CanopyWave", DeploymentID: "canopywave", SortOrder: 17, ChatPreference: 10,
 			RequiresKey: true, CredentialEnv: "CANOPYWAVE_API_KEY",
 			BaseURLEnv: []string{"CANOPYWAVE_BASE_URL"},
 			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "https://inference.canopywave.io/v1",
 			LiveFetcherKey: "canopywave", LiveCatalogKey: "canopywave",
 			ProtocolID: "openai-chat-completions", AdapterID: "canopywave", RuntimeProfileKey: "canopywave",
 		},
 		{
-			ProviderID: "poolside", DisplayName: "Poolside", DeploymentID: "poolside", SortOrder: 19, ChatPreference: 20,
+			ProviderID: "poolside", DisplayName: "Poolside", DeploymentID: "poolside", SortOrder: 18, ChatPreference: 20,
 			RequiresKey: true, CredentialEnv: "POOLSIDE_API_KEY",
 			BaseURLEnv: []string{"POOLSIDE_BASE_URL"},
 			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "https://inference.poolside.ai/v1",
 			LiveFetcherKey: "poolside", LiveCatalogKey: "poolside",
 			ProtocolID: "openai-chat-completions", AdapterID: "poolside", RuntimeProfileKey: "poolside",
 		},
 		{
-			ProviderID: "groq", DisplayName: "Groq", DeploymentID: "groq-direct", SortOrder: 10, ChatPreference: 21,
+			ProviderID: "groq", DisplayName: "Groq", DeploymentID: "groq-direct", SortOrder: 19, ChatPreference: 21,
 			RequiresKey: true, CredentialEnv: "GROQ_API_KEY",
 			BaseURLEnv: []string{"GROQ_BASE_URL"},
 			ProbeKind:  ProbeOpenAIModels, ProbeBaseURL: "https://api.groq.com/openai/v1",
 			LiveFetcherKey: "groq", LiveCatalogKey: "groq",
 			ProtocolID: "openai-chat-completions", AdapterID: "groq", RuntimeProfileKey: "groq",
 		},
 		{
-			ProviderID: "clinepass", DisplayName: "ClinePass", DeploymentID: "clinepass", SortOrder: 6, ChatPreference: 22,
+			ProviderID: "clinepass", DisplayName: "ClinePass", DeploymentID: "clinepass", SortOrder: 20, ChatPreference: 22,
 			RuntimeBaseURL: "https://api.cline.bot/api/v1",
 			RequiresKey:    true, CredentialEnv: "CLINE_API_KEY",
 			BaseURLEnv:     []string{"CLINE_API_BASE"},
@@ -233,7 +233,7 @@ func providerSpecs() []ProviderSpec {
 			ProtocolID: "openai-chat-completions", AdapterID: "clinepass", RuntimeProfileKey: "clinepass",
 		},
 		{
-			ProviderID: "opencodego", DisplayName: "OpenCode Go", DeploymentID: "opencodego", SortOrder: 16, ChatPreference: 13,
+			ProviderID: "opencodego", DisplayName: "OpenCode Go", DeploymentID: "opencodego", SortOrder: 21, ChatPreference: 13,
 			RequiresKey: true, CredentialEnv: "OPENCODEGO_API_KEY",
 			BaseURLEnv:     []string{"OPENCODEGO_BASE_URL"},
 			ProbeKind:      ProbeOpenAIModels,
@@ -244,7 +244,7 @@ func providerSpecs() []ProviderSpec {
 
 		// ── Local ─────────────────────────────────────────────────────────
 		{
-			ProviderID: "ollama", DisplayName: "Ollama", DeploymentID: "ollama-local", SortOrder: 18, ChatPreference: 19,
+			ProviderID: "ollama", DisplayName: "Ollama", DeploymentID: "ollama-local", SortOrder: 22, ChatPreference: 19,
 			RuntimeBaseURL: "http://localhost:11434/v1", RuntimeCredentialEnv: "OLLAMA_API_KEY",
 			RequiresKey: false, CredentialEnv: "OLLAMA_BASE_URL",
 			BaseURLEnv:     []string{"OLLAMA_BASE_URL"},

catalog/v1_test.go

@@ -238,32 +238,6 @@ func TestCapabilitySetFromEntry_ToolsAliasSupportsFunctionCalling(t *testing.T)
 	}
 }
 
-func TestCapabilitySetFromEntry_KeepsContextAndMaxOutput(t *testing.T) {
-	t.Parallel()
-	set := CapabilitySetFromEntry(live.Entry{
-		ContextWindow:   1_048_576,
-		MaxOutput:       131_072,
-		Features:        []string{"tools", "thinking:enabled", "image_input"},
-		ThinkingEnabled: true,
-		ImageInput:      true,
-	})
-	if set.MaxInputTokens != 1_048_576 {
-		t.Fatalf("MaxInputTokens = %d", set.MaxInputTokens)
-	}
-	if set.MaxOutputTokens != 131_072 {
-		t.Fatalf("MaxOutputTokens = %d", set.MaxOutputTokens)
-	}
-	if set.FunctionCalling != CapabilitySupported {
-		t.Fatalf("FunctionCalling = %q", set.FunctionCalling)
-	}
-	if set.ExplicitThinkingBudget != CapabilitySupported {
-		t.Fatalf("ExplicitThinkingBudget = %q", set.ExplicitThinkingBudget)
-	}
-	if set.ImageInput != CapabilitySupported {
-		t.Fatalf("ImageInput = %q", set.ImageInput)
-	}
-}
-
 func TestCapabilitySetFromLegacy_EmptyFeatures(t *testing.T) {
 	t.Parallel()
 	entry := ModelCatalogEntry{ID: "test-model"}

client/adapters/anthropic.go

@@ -159,12 +159,6 @@ func thinkingDisabled() *anthropicThinking {
 }
 
 // resolveThinking builds the thinking config from core.ChatOptions.
-// Explicit ThinkingMode / ThinkingBudgetTokens win. Otherwise ThinkingEnabled
-// maps to Anthropic's documented toggle:
-//   - false → {type:"disabled"}
-//   - true  → {type:"adaptive"} (recommended for current Claude models)
-//
-// See https://platform.claude.com/docs/en/build-with-claude/extended-thinking
 func resolveThinking(opts core.ChatOptions) *anthropicThinking {
 	switch opts.ThinkingMode {
 	case "adaptive":
@@ -178,16 +172,6 @@ func resolveThinking(opts core.ChatOptions) *anthropicThinking {
 		}
 		return thinking
 	default:
-		thinkingEnabled := opts.ThinkingEnabled
-		if thinkingEnabled == nil {
-			thinkingEnabled = opts.GLMThinkingEnabled
-		}
-		if thinkingEnabled != nil {
-			if !*thinkingEnabled {
-				return thinkingDisabled()
-			}
-			return thinkingAdaptive()
-		}
 		// Legacy behavior: if budget > 0, enable with budget
 		return thinkingForBudget(opts.ThinkingBudgetTokens)
 	}