@jeffreycao/copilot-api 2.6.11 → 2.6.13

package/dist/{server-Biyjjk2W.js → server-w3-DhBM0.js}

@@ -1,7 +1,7 @@
-import { A as isResponsesApiWebSocketEnabled, C as getClaudeTokenMultiplier, D as isAlphaSearchCodexPriorityEnabled, E as getUpstreamTransportConfig, O as isMessagesApiEnabled, P as PATHS, S as getClaudeAutoModel, T as getMessageApiWebSearchModel, _ as resolveMappedModel, b as getAlphaSearchModel, d as getModelResponsesApiCompactThreshold$1, f as getReasoningEffortForModel, g as isGpt56OrAbove, h as isContextManagementEnabledForResponses, i as listEnabledProviders, k as isResponsesApiWebSearchEnabled, l as getExtraPromptForModel, m as isContextManagementEnabledForMessages, n as getRawProviderConfig, o as resolveEffectiveProviderType, p as getSmallModel, s as resolveProviderAuthType, t as getProviderConfig, u as getModelMappings, v as setModelMappings, x as getAnthropicApiKey } from "./config-g8kEbc0j.js";
-import { $ as forwardError, B as forwardCodexResponses, C as copilotWebSocketHeaders, E as prepareMessageProxyHeaders, G as createResponsesHttpEventStream, H as requestContext, J as getResponsesStreamTerminalDisposition, K as createResponsesSafeStream, M as compactTextOnlyGuard, N as createAuthMiddleware, O as compactAutoContinuePromptStarts, P as getConfiguredAdminApiKeys, Q as HTTPError, R as CODEX_API_BASE_URL, T as prepareInteractionHeaders, U as resolveTraceId$1, V as generateTraceId, W as fetchUpstreamWithLifecycle, X as createWebSocketUrl, Y as createPooledWebSocketStream, Z as state, b as copilotBaseUrl, d as getUUID, f as isAsyncIterable, g as getCopilotUsage, h as parseUserIdMetadata, j as compactSystemPromptStarts, k as compactMessageSections, l as generateRequestIdFromPayload, m as isResponsesStream, o as setupCodexToken, p as isNullish, q as encodePoolKeyPart, u as getRootSessionId, w as prepareForCompact, x as copilotHeaders, z as buildCodexRequestHeaders } from "./token-SzcMtW4G.js";
+import { B as PATHS, C as resolveMappedModel, D as getAnthropicApiKey, E as getAlphaSearchModel, F as isResponsesApiWebSearchEnabled, I as isResponsesApiWebSocketEnabled, M as getUpstreamTransportConfig, N as isAlphaSearchCodexPriorityEnabled, O as getClaudeAutoModel, P as isMessagesApiEnabled, S as isGpt56OrAbove, _ as getReasoningEffortForModel, b as isContextManagementEnabledForMessages, d as getOpencodeGoModelRecords, g as getModelResponsesApiCompactThreshold$1, h as getModelMappings, i as listEnabledProviders, j as getMessageApiWebSearchModel, k as getClaudeTokenMultiplier, l as getOpencodeGoModelConfig, m as getExtraPromptForModel, n as getRawProviderConfig, o as resolveEffectiveProviderType, s as resolveProviderConfigForModel, t as getProviderConfig, u as getOpencodeGoModelIds, v as getSmallModel, w as setModelMappings, x as isContextManagementEnabledForResponses, y as getSmallModelForProvider } from "./config-B_pC8QXv.js";
+import { $ as forwardError, B as forwardCodexResponses, C as copilotWebSocketHeaders, E as prepareMessageProxyHeaders, G as createResponsesHttpEventStream, H as requestContext, J as getResponsesStreamTerminalDisposition, K as createResponsesSafeStream, M as compactTextOnlyGuard, N as createAuthMiddleware, O as compactAutoContinuePromptStarts, P as getConfiguredAdminApiKeys, Q as HTTPError, R as CODEX_API_BASE_URL, T as prepareInteractionHeaders, U as resolveTraceId$1, V as generateTraceId, W as fetchUpstreamWithLifecycle, X as createWebSocketUrl, Y as createPooledWebSocketStream, Z as state, b as copilotBaseUrl, d as getUUID, f as isAsyncIterable, g as getCopilotUsage, h as parseUserIdMetadata, j as compactSystemPromptStarts, k as compactMessageSections, l as generateRequestIdFromPayload, m as isResponsesStream, o as setupCodexToken, p as isNullish, q as encodePoolKeyPart, u as getRootSessionId, w as prepareForCompact, x as copilotHeaders, z as buildCodexRequestHeaders } from "./token-3SWNt25T.js";
 import { a as isDeferredToolName, c as parseMcpToolSearchSentinel, d as shouldEnableResponsesToolSearch, i as isBridgeToolSearchName, l as resolveBridgeToolSearchName, o as listDeferredToolNames, r as formatToolSearchBridgeArguments, s as normalizeToolSearchBridgeArguments, t as BRIDGE_TOOL_SEARCH_NAME, u as selectDeferredToolsByNames } from "./tool-search-BgFcDbNQ.js";
-import { i as toClientModelId, r as normalizeSdkModelId, t as findEndpointModel } from "./models-C1H-ahM4.js";
+import { i as toClientModelId, r as normalizeSdkModelId, t as findEndpointModel } from "./models-NNzBeGzA.js";
 import consola from "consola";
 import { createHash } from "node:crypto";
 import fs, { createWriteStream, readFileSync, rmSync } from "node:fs";
@@ -550,50 +550,21 @@ const builtinProviderModelRegistry = new class BuiltinProviderModelRegistry {
 					"max"
 				]
 			},
-			"ZHIPU/GLM-5.3": {
-				contextWindow: 1048576,
-				inputModalities: ["text"],
-				maxOutputTokens: 131072,
-				pricing: {
-					cachedInput: 2,
-					input: 8,
-					output: 28
-				},
-				reasoningEfforts: [
-					"low",
-					"high",
-					"max"
-				]
-			},
-			"ZHIPU/GLM-5.3-Flash": {
-				contextWindow: 1048576,
-				inputModalities: ["text", "image"],
-				maxOutputTokens: 131072,
-				pricing: {
-					cachedInput: .23,
-					input: .8,
-					output: 2.8
-				},
-				reasoningEfforts: [
-					"low",
-					"high",
-					"max"
-				]
-			},
-			"ZHIPU/GLM-5.3-FlashX": {
-				contextWindow: 1048576,
+			"deepseek-v4.1-flash": {
+				contextWindow: 1e6,
 				inputModalities: ["text", "image"],
-				maxOutputTokens: 131072,
+				maxOutputTokens: 393216,
 				pricing: {
-					cachedInput: .57,
+					cachedInput: .2,
 					input: 2,
-					output: 7
-				},
-				reasoningEfforts: [
-					"low",
-					"high",
-					"max"
-				]
+					offPeak: {
+						cachedInput: .1,
+						input: 1,
+						output: 4
+					},
+					output: 8,
+					peakWindows: dashscopePeakWindows
+				}
 			},
 			"qwen3.8-max": {
 				contextWindow: 1e6,
@@ -631,22 +602,6 @@ const builtinProviderModelRegistry = new class BuiltinProviderModelRegistry {
 					output: 2.7
 				}
 			},
-			"deepseek-v4.1-flash": {
-				contextWindow: 1e6,
-				inputModalities: ["text", "image"],
-				maxOutputTokens: 393216,
-				pricing: {
-					cachedInput: .2,
-					input: 2,
-					offPeak: {
-						cachedInput: .1,
-						input: 1,
-						output: 4
-					},
-					output: 8,
-					peakWindows: dashscopePeakWindows
-				}
-			},
 			"qwen3.7-plus": {
 				contextWindow: 1e6,
 				inputModalities: ["text", "image"],
@@ -676,95 +631,30 @@ const builtinProviderModelRegistry = new class BuiltinProviderModelRegistry {
 					input: 20,
 					output: 100
 				}
-			}
-		},
-		deepseek: {
-			"deepseek-flash": {
-				contextWindow: 1e6,
-				inputModalities: ["text", "image"],
-				maxOutputTokens: 384e3,
-				pricing: {
-					cachedInput: .04,
-					input: 2,
-					offPeak: {
-						cachedInput: .02,
-						input: 1,
-						output: 4
-					},
-					output: 8,
-					peakWindows: deepseekPeakWindows
-				},
-				reasoningEfforts: [
-					"low",
-					"high",
-					"max"
-				]
 			},
-			"deepseek-v4-pro": {
-				contextWindow: 1e6,
+			"ZHIPU/GLM-5.3": {
+				contextWindow: 1048576,
 				inputModalities: ["text"],
-				maxOutputTokens: 64e3,
+				maxOutputTokens: 131072,
 				pricing: {
-					cachedInput: .3,
-					input: 9,
-					offPeak: {
-						cachedInput: .15,
-						input: 4.5,
-						output: 13.5
-					},
-					output: 27,
-					peakWindows: deepseekPeakWindows
+					cachedInput: 2,
+					input: 8,
+					output: 28
 				},
 				reasoningEfforts: [
 					"low",
 					"high",
 					"max"
 				]
-			}
-		},
-		"opencode-go": {
-			"hy4-preview": {
-				contextWindow: 1024e3,
-				inputModalities: ["text"],
-				maxOutputTokens: 64e3,
-				pricing: {
-					cachedInput: .042,
-					input: .834,
-					output: 2.501
-				},
-				reasoningEfforts: ["high"],
-				reasoningField: "reasoning"
 			},
-			"gpt-6-luna": { pricing: { tiers: [{
-				cacheCreationInput: .125,
-				cachedInput: .01,
-				input: .1,
-				maxInputTokens: 272e3,
-				output: .5
-			}, {
-				cacheCreationInput: .25,
-				cachedInput: .02,
-				input: .2,
-				output: .75
-			}] } },
-			"glm-5.3": {
-				contextWindow: 1e6,
-				inputModalities: ["text"],
-				maxOutputTokens: 64e3,
-				pricing: {
-					cachedInput: .26,
-					input: 1.4,
-					output: 4.4
-				}
-			},
-			"glm-5.3-flash": {
-				contextWindow: 1e6,
+			"ZHIPU/GLM-5.3-Flash": {
+				contextWindow: 1048576,
 				inputModalities: ["text", "image"],
 				maxOutputTokens: 131072,
 				pricing: {
-					cachedInput: .015,
-					input: .075,
-					output: .25
+					cachedInput: .23,
+					input: .8,
+					output: 2.8
 				},
 				reasoningEfforts: [
 					"low",
@@ -772,58 +662,36 @@ const builtinProviderModelRegistry = new class BuiltinProviderModelRegistry {
 					"max"
 				]
 			},
-			"muse-spark-1.3-contributor": {
+			"ZHIPU/GLM-5.3-FlashX": {
 				contextWindow: 1048576,
 				inputModalities: ["text", "image"],
 				maxOutputTokens: 131072,
 				pricing: {
-					cachedInput: .002,
-					input: .1,
-					output: .2
-				},
-				reasoningEfforts: [
-					"minimal",
-					"low",
-					"medium",
-					"high",
-					"xhigh"
-				]
-			},
-			"grok-4.7": {
-				contextWindow: 5e5,
-				defaultReasoningEffort: "high",
-				inputModalities: ["text", "image"],
-				maxOutputTokens: 64e3,
-				pricing: { tiers: [{
-					cachedInput: .5,
+					cachedInput: .57,
 					input: 2,
-					maxInputTokens: 2e5,
-					output: 6
-				}, {
-					cachedInput: 1,
-					input: 4,
-					output: 12
-				}] },
+					output: 7
+				},
 				reasoningEfforts: [
 					"low",
-					"medium",
 					"high",
-					"xhigh"
+					"max"
 				]
-			},
-			"deepseek-v4.1-flash": {
+			}
+		},
+		deepseek: {
+			"deepseek-flash": {
 				contextWindow: 1e6,
 				inputModalities: ["text", "image"],
 				maxOutputTokens: 384e3,
 				pricing: {
-					cachedInput: .006,
-					input: .3,
+					cachedInput: .04,
+					input: 2,
 					offPeak: {
-						cachedInput: .003,
-						input: .15,
-						output: .6
+						cachedInput: .02,
+						input: 1,
+						output: 4
 					},
-					output: 1.2,
+					output: 8,
 					peakWindows: deepseekPeakWindows
 				},
 				reasoningEfforts: [
@@ -837,107 +705,21 @@ const builtinProviderModelRegistry = new class BuiltinProviderModelRegistry {
 				inputModalities: ["text"],
 				maxOutputTokens: 64e3,
 				pricing: {
-					cachedInput: .044,
-					input: 1.32,
+					cachedInput: .3,
+					input: 9,
 					offPeak: {
-						cachedInput: .022,
-						input: .66,
-						output: 1.98
+						cachedInput: .15,
+						input: 4.5,
+						output: 13.5
 					},
-					output: 3.96,
+					output: 27,
 					peakWindows: deepseekPeakWindows
-				}
-			},
-			"kimi-k3": {
-				contextWindow: 1048576,
-				inputModalities: ["text", "image"],
-				maxOutputTokens: 64e3,
-				pricing: {
-					cachedInput: .3,
-					input: 3,
-					output: 15
-				}
-			},
-			"mimo-v2.6-flash": {
-				contextWindow: 1048576,
-				inputModalities: ["text", "image"],
-				maxOutputTokens: 131072,
-				pricing: {
-					cachedInput: .0028,
-					input: .14,
-					output: .28
-				}
-			},
-			"mimo-v2.6-pro": {
-				contextWindow: 1048576,
-				inputModalities: ["text", "image"],
-				maxOutputTokens: 131072,
-				pricing: {
-					cachedInput: .003625,
-					input: .435,
-					output: .87
-				}
-			},
-			"qwen3.7-plus": {
-				contextWindow: 1e6,
-				inputModalities: ["text", "image"],
-				maxOutputTokens: 64e3,
-				pricing: { tiers: [{
-					cacheCreationInput: .5,
-					cachedInput: .04,
-					input: .4,
-					maxInputTokens: 2e5,
-					output: 1.6
-				}, {
-					cacheCreationInput: 1.5,
-					cachedInput: .12,
-					input: 1.2,
-					maxInputTokens: 256e3,
-					output: 4.8
-				}] }
-			},
-			"qwen3.8-max": {
-				contextWindow: 1e6,
-				inputModalities: ["text", "image"],
-				maxOutputTokens: 64e3,
-				pricing: {
-					cacheCreationInput: 2.5,
-					cachedInput: .25,
-					input: 2,
-					output: 6
-				}
-			},
-			"qwen3.8-flash": {
-				contextWindow: 1e6,
-				inputModalities: ["text", "image"],
-				maxOutputTokens: 131072,
-				pricing: {
-					cacheCreationInput: .2,
-					cachedInput: .016,
-					input: .15,
-					output: .47
 				},
 				reasoningEfforts: [
 					"low",
-					"medium",
-					"xhigh"
+					"high",
+					"max"
 				]
-			},
-			"minimax-m3": {
-				contextWindow: 1e6,
-				inputModalities: ["text", "image"],
-				maxOutputTokens: 64e3,
-				pricing: { tiers: [{
-					cachedInput: .06,
-					input: .3,
-					maxInputTokens: 2e5,
-					output: 1.2
-				}, {
-					cachedInput: .12,
-					input: .6,
-					maxInputTokens: 512e3,
-					output: 2.4
-				}] }
 			}
 		},
 		kimi: {
@@ -968,11 +750,15 @@ const builtinProviderModelRegistry = new class BuiltinProviderModelRegistry {
 		this.modelCatalog = BuiltinProviderModelRegistry.catalog;
 	}
 	getModelConfig(providerName, modelName) {
-		const models = this.modelCatalog[this.normalizeKey(providerName)];
+		const provider = this.normalizeKey(providerName);
+		if (provider === "opencode-go") return getOpencodeGoModelConfig(modelName);
+		const models = this.modelCatalog[provider];
 		return models?.[modelName.trim()] ?? models?.[this.normalizeKey(modelName)];
 	}
 	getModelIds(providerName) {
-		return Object.keys(this.modelCatalog[this.normalizeKey(providerName)] ?? {});
+		const provider = this.normalizeKey(providerName);
+		if (provider === "opencode-go") return getOpencodeGoModelIds();
+		return Object.keys(this.modelCatalog[provider] ?? {});
 	}
 	normalizeKey(value) {
 		return value.trim().toLowerCase();
@@ -2356,6 +2142,9 @@ const STRIPPED_RESPONSE_HEADERS = [
 	"transfer-encoding",
 	"upgrade"
 ];
+function resolveProviderEndpointUrl(providerConfig, endpoint) {
+	return `${providerConfig.modelsDevProviderId ? providerConfig.baseUrl.replace(/\/(?:chat\/completions|responses)$/u, "") : `${providerConfig.baseUrl}/v1`}/${endpoint}`;
+}
 function buildProviderUpstreamHeaders(providerConfig, requestHeaders) {
 	const authHeaders = {};
 	if (providerConfig.authType === "x-api-key") authHeaders["x-api-key"] = providerConfig.apiKey;
@@ -2370,6 +2159,7 @@ function buildProviderUpstreamHeaders(providerConfig, requestHeaders) {
 		if (headerValue) headers[headerName] = headerValue;
 	}
 	if (providerConfig.type !== "anthropic") return headers;
+	if (providerConfig.modelsDevProviderId) headers["anthropic-version"] = "2023-06-01";
 	for (const headerName of ANTHROPIC_FORWARDABLE_HEADERS) {
 		const headerValue = requestHeaders.get(headerName);
 		if (headerValue) headers[headerName] = headerValue;
@@ -2407,7 +2197,7 @@ async function forwardProviderMessages(providerConfig, payload, requestHeaders,
 	const headers = buildProviderUpstreamHeaders(providerConfig, requestHeaders);
 	applyOpencodeSessionHeader(providerConfig, headers, resolveOpencodeMessagesSession(payload));
 	const transportConfig = getUpstreamTransportConfig();
-	return await fetchUpstreamWithLifecycle(`${providerConfig.baseUrl}/v1/messages`, {
+	return await fetchUpstreamWithLifecycle(resolveProviderEndpointUrl(providerConfig, "messages"), {
 		method: "POST",
 		headers,
 		body: JSON.stringify(payload)
@@ -2422,7 +2212,7 @@ async function forwardProviderChatCompletions(providerConfig, payload, requestHe
 	const headers = buildProviderUpstreamHeaders(providerConfig, requestHeaders);
 	applyOpencodeSessionHeader(providerConfig, headers, payload.prompt_cache_key?.trim() || void 0);
 	const transportConfig = getUpstreamTransportConfig();
-	return await fetchUpstreamWithLifecycle(`${providerConfig.baseUrl}/v1/chat/completions`, {
+	return await fetchUpstreamWithLifecycle(resolveProviderEndpointUrl(providerConfig, "chat/completions"), {
 		method: "POST",
 		headers,
 		body: JSON.stringify(payload)
@@ -2437,7 +2227,7 @@ async function forwardProviderResponses(providerConfig, payload, requestHeaders,
 	const transportConfig = getUpstreamTransportConfig();
 	const headers = buildProviderUpstreamHeaders(providerConfig, requestHeaders);
 	applyOpencodeSessionHeader(providerConfig, headers, payload.prompt_cache_key?.trim() || void 0);
-	return await fetchUpstreamWithLifecycle(`${providerConfig.baseUrl}/v1/responses`, {
+	return await fetchUpstreamWithLifecycle(resolveProviderEndpointUrl(providerConfig, "responses"), {
 		method: "POST",
 		headers,
 		body: JSON.stringify(payload)
@@ -2449,7 +2239,7 @@ async function forwardProviderResponses(providerConfig, payload, requestHeaders,
 }
 const PROVIDER_MODELS_TIMEOUT_MS = 15e3;
 async function forwardProviderModels(providerConfig, requestHeaders) {
-	return await fetch(`${providerConfig.baseUrl}/v1/models`, {
+	return await fetch(resolveProviderEndpointUrl(providerConfig, "models"), {
 		method: "GET",
 		headers: buildProviderUpstreamHeaders(providerConfig, requestHeaders),
 		signal: AbortSignal.timeout(PROVIDER_MODELS_TIMEOUT_MS)
@@ -2459,7 +2249,7 @@ async function forwardProviderModels(providerConfig, requestHeaders) {
 const PROVIDER_IMAGES_TIMEOUT_MS = 9e5;
 const providerImagesDispatcher = createTimeoutDispatcher(PROVIDER_IMAGES_TIMEOUT_MS);
 function resolveProviderRequestUrl(providerConfig, requestUrl, path) {
-	const upstreamUrl = new URL(`${providerConfig.baseUrl}${path}`);
+	const upstreamUrl = new URL(resolveProviderEndpointUrl(providerConfig, path.replace(/^\/v1\//u, "")));
 	upstreamUrl.search = new URL(requestUrl, "http://localhost").search;
 	return upstreamUrl.toString();
 }
@@ -2831,7 +2621,7 @@ async function resolveRemoteModel(c, request, provider) {
 		if (alphaSearchResponsesDependencies.resolveEffectiveProviderType(providerConfig, model) !== "openai-responses") return invalidRequest$1(c, `Provider '${provider}' does not support the /v1/responses endpoint required for alpha search`);
 		return {
 			model,
-			providerConfig
+			providerConfig: resolveProviderConfigForModel(providerConfig, model)
 		};
 	}
 	if (!alphaSearchResponsesDependencies.findEndpointModel(model)?.supported_endpoints?.includes("/responses")) return invalidRequest$1(c, `Model '${model}' does not support the Copilot Responses endpoint required for alpha search`);
@@ -3262,10 +3052,9 @@ const resolveSupportedReasoningEffort = (requestedEffort, supportedEfforts) => {
 //#region src/routes/provider/utils.ts
 const forwardProviderResponseHeaders = (c, upstreamHeaders) => {
 	const headers = createProviderProxyResponseHeaders(upstreamHeaders);
-	const setCookies = headers.getSetCookie();
 	headers.delete("set-cookie");
+	headers.delete("x-models-etag");
 	for (const [headerName, headerValue] of headers) c.header(headerName, headerValue);
-	for (const setCookie of setCookies) c.header("set-cookie", setCookie, { append: true });
 };
 const applyModelDefaults = (payload, modelConfig) => {
 	payload.temperature ??= modelConfig?.temperature;
@@ -3295,8 +3084,9 @@ const normalizeProviderResponsesReasoningEffort = (payload, providerConfig) => {
 const logger$15 = createHandlerLogger("provider-chat-completions-handler");
 async function handleProviderChatCompletionsForProvider(c, options) {
 	const { payload, provider } = options;
-	const providerConfig = await resolveProviderConfig(provider);
-	if (!providerConfig || resolveEffectiveProviderType(providerConfig, payload.model) !== "openai-compatible") return c.json({ error: {
+	const configuredProvider = await resolveProviderConfig(provider);
+	const providerConfig = configuredProvider && resolveProviderConfigForModel(configuredProvider, payload.model);
+	if (!providerConfig || providerConfig.type !== "openai-compatible") return c.json({ error: {
 		message: `Provider '${provider}' does not support the /v1/chat/completions endpoint`,
 		type: "invalid_request_error"
 	} }, 400);
@@ -5571,6 +5361,7 @@ const translateAnthropicMessagesToResponsesPayload = (payload, subagentAgentId)
 	};
 	if (translatedTools && translatedTools.length > 0) responsesPayload.tool_choice = toolChoice;
 	if (hasOriginalTools) responsesPayload.prompt_cache_key = promptCacheKey;
+	if (responsesPayload.model.includes("grok")) delete responsesPayload.metadata;
 	return responsesPayload;
 };
 const encodeCompactionCarrierSignature = (compaction) => {
@@ -8029,10 +7820,6 @@ function getModels() {
 //#region src/routes/provider/messages/handler.ts
 const logger$10 = createHandlerLogger("provider-messages-handler");
 const providerMessagesHandlerDependencies = { resolveProviderConfig };
-const resolveOverrideProviderAuthType = (providerConfig, effectiveType) => {
-	if (providerConfig.authType === "azure-entra" || providerConfig.authType === "oauth2") return providerConfig.authType;
-	return resolveProviderAuthType(providerConfig.name, void 0, effectiveType);
-};
 async function handleProviderMessages(c) {
 	const provider = c.req.param("provider");
 	const payload = await c.req.json();
@@ -8048,14 +7835,15 @@ async function handleProviderMessages(c) {
 }
 async function handleProviderMessagesForProvider(c, options) {
 	const { payload, provider, usageEndpoint } = options;
-	const providerConfig = await providerMessagesHandlerDependencies.resolveProviderConfig(provider);
-	if (!providerConfig) return c.json({ error: {
+	const configuredProvider = await providerMessagesHandlerDependencies.resolveProviderConfig(provider);
+	if (!configuredProvider) return c.json({ error: {
 		message: `Provider '${provider}' not found or disabled`,
 		type: "invalid_request_error"
 	} }, 404);
 	try {
+		const providerConfig = resolveProviderConfigForModel(configuredProvider, payload.model);
 		const modelConfig = providerConfig.models?.[payload.model];
-		const effectiveType = resolveEffectiveProviderType(providerConfig, payload.model);
+		const effectiveType = providerConfig.type;
 		debugJson(logger$10, "provider.messages.request", {
 			payload,
 			provider
@@ -8096,11 +7884,7 @@ async function handleProviderMessagesForProvider(c, options) {
 			payload,
 			provider
 		});
-		const upstreamResponse = await forwardProviderMessages(effectiveType === providerConfig.type ? providerConfig : {
-			...providerConfig,
-			type: effectiveType,
-			authType: resolveOverrideProviderAuthType(providerConfig, effectiveType)
-		}, payload, c.req.raw.headers, { clientSignal: c.req.raw.signal });
+		const upstreamResponse = await forwardProviderMessages(providerConfig, payload, c.req.raw.headers, { clientSignal: c.req.raw.signal });
 		if (!upstreamResponse.ok) {
 			logger$10.error("Failed to create responses", upstreamResponse);
 			throw new HTTPError("Failed to create responses", upstreamResponse);
@@ -9133,7 +8917,11 @@ async function handleCompletionPayload(c, anthropicPayload, dispatchOptions = {}
 	if (!state.tokenBasedBilling && !shouldUseClaudeAutoModel) {
 		const tools = anthropicPayload.tools;
 		const noTools = !tools || tools.length === 0;
-		if (anthropicBeta && noTools && compactType === 0) anthropicPayload.model = getSmallModel();
+		if (anthropicBeta && noTools && compactType === 0) {
+			const smallModel = getSmallModel();
+			consola.debug(`Claude Code warmup small model: ${anthropicPayload.model} -> ${smallModel}`);
+			anthropicPayload.model = smallModel;
+		}
 	}
 	if (compactType) logger$9.debug("Compact request type:", compactType);
 	if (!state.tokenBasedBilling) {
@@ -10874,6 +10662,7 @@ function normalizeProviderModels(provider, models) {
 	return models.map((model) => normalizeProviderModel(provider, model)).filter((model) => model !== null);
 }
 async function getProviderModelRecords(providerConfig, requestHeaders) {
+	if (providerConfig.name === "opencode-go") return getOpencodeGoModelRecords();
 	try {
 		const response = await forwardProviderModels(providerConfig, requestHeaders);
 		if (!response.ok) return getFallbackProviderModelRecords(providerConfig.name, "non_ok", { statusCode: response.status });
@@ -11200,6 +10989,18 @@ providerModelRoutes.get("/", async (c) => {
 				has_more: false
 			});
 		}
+		if (providerConfig.name === "opencode-go") {
+			const models = getOpencodeGoModelRecords();
+			if (models.length === 0) return c.json({ error: {
+				message: "OpenCode Go model catalog is unavailable",
+				type: "service_unavailable"
+			} }, 503);
+			return c.json({
+				object: "list",
+				data: models,
+				has_more: false
+			});
+		}
 		const upstreamResponse = await forwardProviderModels(providerConfig, c.req.raw.headers);
 		logger$4.debug("provider.models.response", {
 			provider,
@@ -12149,28 +11950,46 @@ function isAnthropicResponse(value) {
 	return typeof value === "object" && value !== null && "type" in value && value.type === "message" && "content" in value && Array.isArray(value.content);
 }
 //#endregion
+//#region src/routes/responses/task-title.ts
+const TASK_TITLE_PROMPT_PREFIX = "Generate a concise, single-line task title of at most 36 characters and under five words where possible.";
+const taskTitleDependencies = { getSmallModelForProvider };
+const getCodexTaskTitleModel = (userAgent, input, provider) => {
+	if (!isCodexUserAgent(userAgent) || !Array.isArray(input)) return void 0;
+	if (!input.slice(-2).some((item) => {
+		const message = item;
+		return message?.role === "user" && Array.isArray(message.content) && message.content.some((part) => part?.type === "input_text" && typeof part.text === "string" && part.text.startsWith(TASK_TITLE_PROMPT_PREFIX));
+	})) return void 0;
+	const model = taskTitleDependencies.getSmallModelForProvider(provider);
+	if (model) consola.debug(`Codex task title small model: ${provider} -> ${model}`);
+	return model;
+};
+//#endregion
 //#region src/routes/provider/responses/handler.ts
 const logger$2 = createHandlerLogger("provider-responses-handler");
 const providerResponsesHandlerDependencies = { resolveProviderConfig };
 async function handleProviderResponsesForProvider(c, options) {
 	const { payload, provider } = options;
+	const taskTitleModel = getCodexTaskTitleModel(c.req.header("user-agent"), payload.input, provider);
+	if (taskTitleModel) payload.model = taskTitleModel;
+	const publicModel = taskTitleModel ?? options.publicModel ?? payload.model;
 	debugJson(logger$2, "Responses request payload:", {
 		payload,
 		provider
 	});
-	const providerConfig = await providerResponsesHandlerDependencies.resolveProviderConfig(provider);
+	const configuredProvider = await providerResponsesHandlerDependencies.resolveProviderConfig(provider);
+	const providerConfig = configuredProvider && resolveProviderConfigForModel(configuredProvider, payload.model);
 	if (!providerConfig) return c.json({ error: {
 		message: `Provider '${provider}' does not support the /v1/responses endpoint`,
 		type: "invalid_request_error"
 	} }, 400);
-	const effectiveType = resolveEffectiveProviderType(providerConfig, payload.model);
+	const effectiveType = providerConfig.type;
 	const normalizedReasoningEffort = normalizeProviderResponsesReasoningEffort(payload, providerConfig);
 	if (normalizedReasoningEffort) logger$2.debug(`Normalized reasoning effort from ${normalizedReasoningEffort.from} to ${normalizedReasoningEffort.to} based on the provider model configuration`);
 	if (shouldFallbackToMessages$1(c, payload.model, effectiveType)) {
 		filterReasoningForTransport(payload, true);
 		return await handleResponsesViaMessages(c, {
 			payload,
-			publicModel: options.publicModel ?? payload.model,
+			publicModel,
 			targetModel: `${provider}/${payload.model}`
 		});
 	}
@@ -12382,6 +12201,9 @@ const handleResponses = async (c) => {
 			publicModel: requestedModel
 		});
 	}
+	const taskTitleModel = getCodexTaskTitleModel(c.req.header("user-agent"), payload.input, "copilot");
+	if (taskTitleModel) payload.model = taskTitleModel;
+	const publicModel = taskTitleModel ?? requestedModel;
 	debugJson(logger$1, "Responses request payload:", payload);
 	const subagentMarker = getCodexResponsesSubagentMarker(c);
 	if (subagentMarker) debugJson(logger$1, "Detected Codex subagent headers:", subagentMarker);
@@ -12400,7 +12222,7 @@ const handleResponses = async (c) => {
 		filterReasoningForTransport(payload, true);
 		return await handleResponsesViaMessages(c, {
 			payload,
-			publicModel: requestedModel,
+			publicModel,
 			targetModel: payload.model,
 			subagentMarker,
 			requestId,
@@ -12693,4 +12515,4 @@ createServer();
 //#endregion
 export { createServer };
 
-//# sourceMappingURL=server-Biyjjk2W.js.map
+//# sourceMappingURL=server-w3-DhBM0.js.map