@gajae-code/ai 0.11.8 → 0.11.9

package/CHANGELOG.md

@@ -2,12 +2,21 @@
 
 ## [Unreleased]
 
+## [0.11.9] - 2026-07-24
+### Fixed
+
+- Kimi Code now allows one continuous 300-second first-event wait before aborting, while preserving explicit caller and environment timeout overrides and the existing inter-event idle timeout.
+
+### Added
+
+- Added first-class support for **OpenGateway by Sionic AI**, an OpenAI-compatible gateway. Registers the `opengateway` provider descriptor, `/login` OAuth entry (API-key paste validated against `https://apis.opengateway.ai/v1/models`), `OPENGATEWAY_API_KEY` environment resolution, and bundled `models.json` seed models. Models are discovered dynamically from the OpenAI-compatible `/v1/models` endpoint (base URL `https://apis.opengateway.ai/v1`).
+
 ## [0.11.8] - 2026-07-23
 
 ### Fixed
 
 - OpenAI Responses / Codex native history replay no longer submits missing resident-image placeholders as `input_image.image_url`. Invalid values (including `[Session resident imageUrl blob missing: …]`) are dropped, or retained as `file_id`-only parts when a non-empty `file_id` is present, so a single unavailable historical image cannot brick `/retry` (#2924).
-- Raised the first-event stream timeout floor to five minutes for `alibaba-token-plan` models using the OpenAI Completions and Responses APIs, while preserving caller and environment overrides and the existing inter-event idle timeout.
+- Raised the first-event stream timeout floor to five minutes for `alibaba-token-plan` models at both the OpenAI provider and outer lazy-stream watchdogs, while preserving caller and environment overrides and the existing inter-event idle timeout.
 - OAuth refresh peer-rotation recovery now runs before failure classification instead of only on the definitive-failure path, and the definitive matcher recognizes the "grant is invalid" phrasing. Providers whose invalid-grant response does not contain the literal `invalid_grant` (e.g. Kimi's 400 "The provided authorization grant is invalid") previously had rotation races misclassified as transient, temp-blocking a healthy credential for five minutes on every race; with Kimi's ~12-minute access tokens and multiple processes sharing the credential store this surfaced as repeated logouts. Genuine revocations are now disabled with a cause instead of looping temp-blocks.
 - Anthropic 400 `Invalid \`signature\` in \`thinking\` block` responses now trigger the one-shot thinking replay repair instead of failing the turn. The existing repair matcher only recognized the "latest assistant message ... cannot be modified" wording, so the signature-validation variant — which can cite a `thinking`/`redacted_thinking` block anywhere in the replayed history (e.g. after compaction/pruning rewrote an earlier turn) — was treated as a fatal request error. The retry now rebuilds the request with thinking blocks dropped from every replayed assistant message (`repairAllAssistantThinking`), while the latest-message mutation variant keeps the targeted latest-only repair.
 

package/README.md

@@ -69,6 +69,7 @@ Unified LLM API with automatic model discovery, provider configuration, token an
 - **MiniMax Coding Plan** (requires `MINIMAX_CODE_API_KEY` or `MINIMAX_CODE_CN_API_KEY`)
 - **Xiaomi MiMo** (requires `XIAOMI_API_KEY`)
 - **ZenMux** (requires `ZENMUX_API_KEY`)
+- **OpenGateway by Sionic AI** (requires `OPENGATEWAY_API_KEY`)
 - **Qwen Portal** (supports `QWEN_OAUTH_TOKEN` or `QWEN_PORTAL_API_KEY`)
 - **Cloudflare AI Gateway** (requires `CLOUDFLARE_AI_GATEWAY_API_KEY` and provider-specific gateway base URL)
 - **Ollama** (local OpenAI-compatible runtime; optional `OLLAMA_API_KEY`)
@@ -954,6 +955,7 @@ In Node.js environments, you can set environment variables to avoid passing API
 | MiniMax Code   | `MINIMAX_CODE_API_KEY` (international) or `MINIMAX_CODE_CN_API_KEY` (China) |
 | Xiaomi MiMo    | `XIAOMI_API_KEY`                                                             |
 | ZenMux         | `ZENMUX_API_KEY`                                                             |
+| OpenGateway    | `OPENGATEWAY_API_KEY`                                                        |
 | vLLM           | `VLLM_API_KEY`                                                               |
 | Cloudflare AI Gateway | `CLOUDFLARE_AI_GATEWAY_API_KEY`                                      |
 | GitHub Copilot | `COPILOT_GITHUB_TOKEN` or `GH_TOKEN` or `GITHUB_TOKEN`                      |
@@ -977,6 +979,7 @@ Provider endpoint defaults for the current OpenAI-compatible integrations:
 - Xiaomi MiMo: `https://api.xiaomimimo.com/anthropic`
 - ZenMux (OpenAI): `https://zenmux.ai/api/v1`
 - ZenMux (Anthropic models): `https://zenmux.ai/api/anthropic`
+- OpenGateway by Sionic AI: `https://apis.opengateway.ai/v1`
 - vLLM: `http://127.0.0.1:8000/v1`
 - Ollama: local OpenAI-compatible runtime (`http://127.0.0.1:11434/v1`)
 - Ollama Cloud: native Ollama API host (`https://ollama.com/api`, configured here as base URL `https://ollama.com`)

package/dist/types/provider-models/openai-compat.d.ts

@@ -111,6 +111,16 @@ export interface ZenMuxModelManagerConfig {
     baseUrl?: string;
 }
 export declare function zenmuxModelManagerOptions(config?: ZenMuxModelManagerConfig): ModelManagerOptions<Api>;
+export interface OpenGatewayModelManagerConfig {
+    apiKey?: string;
+    baseUrl?: string;
+}
+/**
+ * OpenGateway by Sionic AI — an OpenAI-compatible gateway that fronts OpenAI,
+ * Anthropic, and Google models behind one API key. Models are discovered from
+ * the OpenAI-compatible `/v1/models` endpoint.
+ */
+export declare function opengatewayModelManagerOptions(config?: OpenGatewayModelManagerConfig): ModelManagerOptions<"openai-completions">;
 export interface KiloModelManagerConfig {
     apiKey?: string;
     baseUrl?: string;

package/dist/types/providers/register-builtins.d.ts

@@ -17,6 +17,14 @@ interface BedrockProviderModule {
     streamBedrock: (model: Model<"bedrock-converse-stream">, context: Context, options: BedrockOptions) => AssistantMessageEventStream;
 }
 export declare function setBedrockProviderModule(module: BedrockProviderModule): void;
+/**
+ * Resolves the first-event timeout fallback for the outer lazy-stream watchdog.
+ * A configured wrapper-specific fallback (from `LazyStreamLimits`) always wins;
+ * otherwise providers known to have slow first events get a five-minute floor
+ * matching their inner provider-level override. Returns `undefined` for
+ * providers that should use the shared default.
+ */
+export declare function resolveLazyStreamFirstEventFallbackMs(provider: string, configuredFallbackMs?: number): number | undefined;
 export declare const streamAnthropic: (model: Model<"anthropic-messages">, context: Context, options: OptionsForApi<"anthropic-messages">) => EventStreamImpl;
 export declare const streamAzureOpenAIResponses: (model: Model<"azure-openai-responses">, context: Context, options: OptionsForApi<"azure-openai-responses">) => EventStreamImpl;
 export declare const streamGoogle: (model: Model<"google-generative-ai">, context: Context, options: OptionsForApi<"google-generative-ai">) => EventStreamImpl;

package/dist/types/types.d.ts

@@ -51,7 +51,7 @@ export interface ThinkingConfig {
     /** Provider-specific transport used to encode the selected effort. */
     mode: ThinkingControlMode;
 }
-export type KnownProvider = "alibaba-token-plan" | "amazon-bedrock" | "azure-openai" | "anthropic" | "google" | "google-gemini-cli" | "google-antigravity" | "google-vertex" | "openai" | "openai-codex" | "kimi-code" | "minimax-code" | "minimax-code-cn" | "github-copilot" | "fireworks" | "firepass" | "fugu" | "gitlab-duo" | "cursor" | "deepseek" | "deepinfra" | "xai" | "groq" | "cerebras" | "openrouter" | "kilo" | "vercel-ai-gateway" | "zai" | "glm-zcode" | "mistral" | "minimax" | "opencode-go" | "opencode-zen" | "synthetic" | "cloudflare-ai-gateway" | "huggingface" | "litellm" | "moonshot" | "nvidia" | "nanogpt" | "ollama" | "ollama-cloud" | "qianfan" | "qwen-portal" | "together" | "venice" | "vllm" | "xiaomi" | "xiaomi-token-plan-sgp" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "zenmux" | "lm-studio";
+export type KnownProvider = "alibaba-token-plan" | "amazon-bedrock" | "azure-openai" | "anthropic" | "google" | "google-gemini-cli" | "google-antigravity" | "google-vertex" | "openai" | "openai-codex" | "kimi-code" | "minimax-code" | "minimax-code-cn" | "github-copilot" | "fireworks" | "firepass" | "fugu" | "gitlab-duo" | "cursor" | "deepseek" | "deepinfra" | "xai" | "groq" | "cerebras" | "openrouter" | "kilo" | "vercel-ai-gateway" | "zai" | "glm-zcode" | "mistral" | "minimax" | "opencode-go" | "opencode-zen" | "opengateway" | "synthetic" | "cloudflare-ai-gateway" | "huggingface" | "litellm" | "moonshot" | "nvidia" | "nanogpt" | "ollama" | "ollama-cloud" | "qianfan" | "qwen-portal" | "together" | "venice" | "vllm" | "xiaomi" | "xiaomi-token-plan-sgp" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "zenmux" | "lm-studio";
 export type Provider = KnownProvider | string;
 import type { Effort } from "./model-thinking";
 /** Token budgets for each thinking level (token-based providers only) */

package/dist/types/utils/idle-iterator.d.ts

@@ -1,3 +1,4 @@
+export declare function getProviderFirstEventTimeoutFallbackMs(provider: string): number | undefined;
 /**
  * Returns the idle timeout used for provider streaming transports.
  *

package/dist/types/utils/oauth/opengateway.d.ts

@@ -0,0 +1 @@
+export declare const loginOpenGateway: (options: import("./types").OAuthController) => Promise<string>;

package/dist/types/utils/oauth/types.d.ts

@@ -7,7 +7,7 @@ export type OAuthCredentials = {
     email?: string;
     accountId?: string;
 };
-export type OAuthProvider = "alibaba-token-plan" | "anthropic" | "cerebras" | "cloudflare-ai-gateway" | "cursor" | "deepseek" | "deepinfra" | "fireworks" | "firepass" | "fugu" | "github-copilot" | "google-gemini-cli" | "google-antigravity" | "gitlab-duo" | "huggingface" | "kimi-code" | "kilo" | "kagi" | "litellm" | "lm-studio" | "minimax-code" | "minimax-code-cn" | "moonshot" | "nvidia" | "nanogpt" | "ollama" | "ollama-cloud" | "openai-codex" | "openai-codex-device" | "opencode-go" | "opencode-zen" | "parallel" | "perplexity" | "qianfan" | "qwen-portal" | "synthetic" | "tavily" | "together" | "venice" | "vercel-ai-gateway" | "vllm" | "xai" | "glm-zcode" | "xiaomi" | "xiaomi-token-plan-sgp" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "zenmux" | "zai";
+export type OAuthProvider = "alibaba-token-plan" | "anthropic" | "cerebras" | "cloudflare-ai-gateway" | "cursor" | "deepseek" | "deepinfra" | "fireworks" | "firepass" | "fugu" | "github-copilot" | "google-gemini-cli" | "google-antigravity" | "gitlab-duo" | "huggingface" | "kimi-code" | "kilo" | "kagi" | "litellm" | "lm-studio" | "minimax-code" | "minimax-code-cn" | "moonshot" | "nvidia" | "nanogpt" | "ollama" | "ollama-cloud" | "openai-codex" | "openai-codex-device" | "opencode-go" | "opencode-zen" | "opengateway" | "parallel" | "perplexity" | "qianfan" | "qwen-portal" | "synthetic" | "tavily" | "together" | "venice" | "vercel-ai-gateway" | "vllm" | "xai" | "glm-zcode" | "xiaomi" | "xiaomi-token-plan-sgp" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "zenmux" | "zai";
 export type OAuthProviderId = OAuthProvider | (string & {});
 export type OAuthPrompt = {
     message: string;

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@gajae-code/ai",
-	"version": "0.11.8",
+	"version": "0.11.9",
 	"description": "Unified LLM API with automatic model discovery and provider configuration",
 	"homepage": "https://gajae-code.com",
 	"author": "Yeachan-Heo and Gajae Code Contributors",
@@ -40,7 +40,7 @@
 	"dependencies": {
 		"@anthropic-ai/sdk": "^0.94.0",
 		"@bufbuild/protobuf": "^2.12.0",
-		"@gajae-code/utils": "0.11.8",
+		"@gajae-code/utils": "0.11.9",
 		"openai": "^6.36.0",
 		"partial-json": "^0.1.7",
 		"zod": "4.4.3"

package/src/auth-storage.ts

@@ -1998,6 +1998,12 @@ export class AuthStorage {
 				await saveApiKeyCredential(apiKey);
 				return;
 			}
+			case "opengateway": {
+				const { loginOpenGateway } = await import("./utils/oauth/opengateway");
+				const apiKey = await loginOpenGateway(ctrl);
+				await saveApiKeyCredential(apiKey);
+				return;
+			}
 			default: {
 				const customProvider = getOAuthProvider(provider);
 				if (!customProvider) {

package/src/cli.ts

@@ -118,6 +118,7 @@ Providers:
   minimax-code-cn   MiniMax Coding Plan (China)
   cursor            Cursor (Anthropic, GPT, etc.)
   zenmux            ZenMux
+  opengateway       OpenGateway by Sionic AI
   ollama-cloud      Ollama Cloud
 
 Examples:

package/src/models.json

@@ -61333,6 +61333,78 @@
 			"maxTokens": 131072
 		}
 	},
+	"opengateway": {
+		"openai/gpt-4o": {
+			"id": "openai/gpt-4o",
+			"name": "GPT-4o (OpenGateway)",
+			"api": "openai-completions",
+			"provider": "opengateway",
+			"baseUrl": "https://apis.opengateway.ai/v1",
+			"reasoning": false,
+			"input": [
+				"text",
+				"image"
+			],
+			"cost": {
+				"input": 2.5,
+				"output": 10,
+				"cacheRead": 1.25,
+				"cacheWrite": 0
+			},
+			"contextWindow": 128000,
+			"maxTokens": 16384
+		},
+		"anthropic/claude-sonnet-4-5": {
+			"id": "anthropic/claude-sonnet-4-5",
+			"name": "Anthropic Sonnet 4.5 (OpenGateway)",
+			"api": "openai-completions",
+			"provider": "opengateway",
+			"baseUrl": "https://apis.opengateway.ai/v1",
+			"reasoning": true,
+			"input": [
+				"text",
+				"image"
+			],
+			"cost": {
+				"input": 3,
+				"output": 15,
+				"cacheRead": 0.3,
+				"cacheWrite": 3.75
+			},
+			"contextWindow": 200000,
+			"maxTokens": 64000,
+			"thinking": {
+				"mode": "effort",
+				"minLevel": "minimal",
+				"maxLevel": "high"
+			}
+		},
+		"google/gemini-2.5-pro": {
+			"id": "google/gemini-2.5-pro",
+			"name": "Gemini 2.5 Pro (OpenGateway)",
+			"api": "openai-completions",
+			"provider": "opengateway",
+			"baseUrl": "https://apis.opengateway.ai/v1",
+			"reasoning": true,
+			"input": [
+				"text",
+				"image"
+			],
+			"cost": {
+				"input": 1.25,
+				"output": 10,
+				"cacheRead": 0.31,
+				"cacheWrite": 0
+			},
+			"contextWindow": 1048576,
+			"maxTokens": 65536,
+			"thinking": {
+				"mode": "effort",
+				"minLevel": "minimal",
+				"maxLevel": "high"
+			}
+		}
+	},
 	"openrouter": {
 		"~anthropic/claude-fable-latest": {
 			"id": "~anthropic/claude-fable-latest",

package/src/provider-models/descriptors.ts

@@ -33,6 +33,7 @@ import {
 	openaiModelManagerOptions,
 	opencodeGoModelManagerOptions,
 	opencodeZenModelManagerOptions,
+	opengatewayModelManagerOptions,
 	openrouterModelManagerOptions,
 	qianfanModelManagerOptions,
 	qwenPortalModelManagerOptions,
@@ -312,6 +313,12 @@ export const PROVIDER_DESCRIPTORS: readonly ProviderDescriptor[] = [
 		config => zenmuxModelManagerOptions(config),
 		catalog("ZenMux", ["ZENMUX_API_KEY"]),
 	),
+	catalogDescriptor(
+		"opengateway",
+		"openai/gpt-4o",
+		config => opengatewayModelManagerOptions(config),
+		catalog("OpenGateway by Sionic AI", ["OPENGATEWAY_API_KEY"]),
+	),
 	catalogDescriptor("zai", "glm-5.2", config => zaiModelManagerOptions(config), catalog("zAI", ["ZAI_API_KEY"])),
 	catalogDescriptor(
 		"glm-zcode",

package/src/provider-models/openai-compat.ts

@@ -1099,6 +1099,26 @@ export function zenmuxModelManagerOptions(config?: ZenMuxModelManagerConfig): Mo
 	};
 }
 
+// ---------------------------------------------------------------------------
+// 10.5.1 OpenGateway by Sionic AI
+// ---------------------------------------------------------------------------
+
+export interface OpenGatewayModelManagerConfig {
+	apiKey?: string;
+	baseUrl?: string;
+}
+
+/**
+ * OpenGateway by Sionic AI — an OpenAI-compatible gateway that fronts OpenAI,
+ * Anthropic, and Google models behind one API key. Models are discovered from
+ * the OpenAI-compatible `/v1/models` endpoint.
+ */
+export function opengatewayModelManagerOptions(
+	config?: OpenGatewayModelManagerConfig,
+): ModelManagerOptions<"openai-completions"> {
+	return createSimpleOpenAICompletionsOptions("opengateway", "https://apis.opengateway.ai/v1", config);
+}
+
 // ---------------------------------------------------------------------------
 // 10.6 Kilo Gateway
 // ---------------------------------------------------------------------------

package/src/providers/anthropic.ts

@@ -60,7 +60,12 @@ import { AssistantMessageEventStream } from "../utils/event-stream";
 import { transportFailureFacts } from "../utils/fallback-transport";
 import { isFoundryEnabled } from "../utils/foundry";
 import { finalizeErrorMessage, type RawHttpRequestDump, rewriteCopilotError } from "../utils/http-inspector";
-import { getStreamFirstEventTimeoutMs, getStreamIdleTimeoutMs, iterateWithIdleTimeout } from "../utils/idle-iterator";
+import {
+	getProviderFirstEventTimeoutFallbackMs,
+	getStreamFirstEventTimeoutMs,
+	getStreamIdleTimeoutMs,
+	iterateWithIdleTimeout,
+} from "../utils/idle-iterator";
 import { parseJsonWithRepair, parseStreamingJson } from "../utils/json-parse";
 import { parseGitHubCopilotApiKey } from "../utils/oauth/github-copilot";
 import { notifyProviderResponse } from "../utils/provider-response";
@@ -1416,7 +1421,9 @@ export const streamAnthropic: StreamFunction<"anthropic-messages"> = (
 				firstTokenTime = undefined;
 			};
 			const idleTimeoutMs = options?.streamIdleTimeoutMs ?? getStreamIdleTimeoutMs();
-			const firstEventTimeoutMs = options?.streamFirstEventTimeoutMs ?? getStreamFirstEventTimeoutMs(idleTimeoutMs);
+			const firstEventFallbackMs = getProviderFirstEventTimeoutFallbackMs(model.provider);
+			const firstEventTimeoutMs =
+				options?.streamFirstEventTimeoutMs ?? getStreamFirstEventTimeoutMs(idleTimeoutMs, firstEventFallbackMs);
 			stream.push({ type: "start", partial: output });
 			// Retry loop for transient errors from the stream.
 			// Provider-level transport/rate-limit failures: only before any streamed content starts.

package/src/providers/openai-anthropic-shim.ts

@@ -89,6 +89,8 @@ export function streamOpenAIAnthropicShim(
 					onResponse: options?.onResponse,
 					onSseEvent: options?.onSseEvent,
 					fetch: options?.fetch,
+					streamIdleTimeoutMs: options?.streamIdleTimeoutMs,
+					streamFirstEventTimeoutMs: options?.streamFirstEventTimeoutMs,
 					thinkingEnabled,
 					thinkingBudgetTokens: thinkingBudget,
 				});
@@ -118,6 +120,8 @@ export function streamOpenAIAnthropicShim(
 					onResponse: options?.onResponse,
 					onSseEvent: options?.onSseEvent,
 					fetch: options?.fetch,
+					streamIdleTimeoutMs: options?.streamIdleTimeoutMs,
+					streamFirstEventTimeoutMs: options?.streamFirstEventTimeoutMs,
 					reasoning: reasoningEffort,
 				});
 

package/src/providers/openai-completions.ts

@@ -49,6 +49,7 @@ import {
 import {
 	createWatchdog,
 	getOpenAIStreamIdleTimeoutMs,
+	getProviderFirstEventTimeoutFallbackMs,
 	getStreamFirstEventTimeoutMs,
 	iterateWithIdleTimeout,
 } from "../utils/idle-iterator";
@@ -427,6 +428,7 @@ function getTrailingPartialDeepseekToken(text: string): string {
 }
 
 const ALIBABA_TOKEN_PLAN_FIRST_EVENT_TIMEOUT_MS = 300_000;
+
 const OPENAI_COMPLETIONS_FIRST_EVENT_TIMEOUT_MESSAGE =
 	"OpenAI completions stream timed out while waiting for the first event";
 
@@ -564,7 +566,9 @@ export const streamOpenAICompletions: StreamFunction<"openai-completions"> = (
 				}
 			}
 			const firstEventFallbackMs =
-				model.provider === "alibaba-token-plan" ? ALIBABA_TOKEN_PLAN_FIRST_EVENT_TIMEOUT_MS : undefined;
+				model.provider === "alibaba-token-plan"
+					? ALIBABA_TOKEN_PLAN_FIRST_EVENT_TIMEOUT_MS
+					: getProviderFirstEventTimeoutFallbackMs(model.provider);
 			const firstEventWatchdog = createWatchdog(
 				options?.streamFirstEventTimeoutMs ?? getStreamFirstEventTimeoutMs(idleTimeoutMs, firstEventFallbackMs),
 				() => abortTracker.abortLocally(firstEventTimeoutAbortError),

package/src/providers/register-builtins.ts

@@ -190,6 +190,22 @@ interface LazyStreamLimits {
 const GOOGLE_GEMINI_CLI_LAZY_STREAM_LIMITS: LazyStreamLimits = {
 	defaultFirstEventTimeoutMs: 300_000,
 };
+const SLOW_FIRST_EVENT_PROVIDERS = new Set(["alibaba-token-plan", "kimi-code"]);
+
+/**
+ * Resolves the first-event timeout fallback for the outer lazy-stream watchdog.
+ * A configured wrapper-specific fallback (from `LazyStreamLimits`) always wins;
+ * otherwise providers known to have slow first events get a five-minute floor
+ * matching their inner provider-level override. Returns `undefined` for
+ * providers that should use the shared default.
+ */
+export function resolveLazyStreamFirstEventFallbackMs(
+	provider: string,
+	configuredFallbackMs?: number,
+): number | undefined {
+	if (configuredFallbackMs !== undefined) return configuredFallbackMs;
+	return SLOW_FIRST_EVENT_PROVIDERS.has(provider) ? 300_000 : undefined;
+}
 
 function forwardStream<TApi extends Api>(
 	target: EventStreamImpl,
@@ -202,11 +218,14 @@ function forwardStream<TApi extends Api>(
 	(async () => {
 		try {
 			const idleTimeoutMs = options.streamIdleTimeoutMs ?? getStreamIdleTimeoutMs(limits?.defaultIdleTimeoutMs);
+			const firstEventFallbackMs = resolveLazyStreamFirstEventFallbackMs(
+				model.provider,
+				limits?.defaultFirstEventTimeoutMs,
+			);
 			const watchedSource = iterateWithIdleTimeout(source, {
 				idleTimeoutMs,
 				firstItemTimeoutMs:
-					options.streamFirstEventTimeoutMs ??
-					getStreamFirstEventTimeoutMs(idleTimeoutMs, limits?.defaultFirstEventTimeoutMs),
+					options.streamFirstEventTimeoutMs ?? getStreamFirstEventTimeoutMs(idleTimeoutMs, firstEventFallbackMs),
 				errorMessage: LAZY_STREAM_IDLE_TIMEOUT_ERROR,
 				firstItemErrorMessage: LAZY_STREAM_FIRST_EVENT_TIMEOUT_ERROR,
 				onIdle: () => abortTracker.abortLocally(new Error(LAZY_STREAM_IDLE_TIMEOUT_ERROR)),

package/src/stream.ts

@@ -162,6 +162,7 @@ const serviceProviderMap: Record<string, KeyResolver> = {
 	"qwen-portal": () => $pickCredentialEnv("QWEN_OAUTH_TOKEN", "QWEN_PORTAL_API_KEY"),
 	together: "TOGETHER_API_KEY",
 	zenmux: "ZENMUX_API_KEY",
+	opengateway: "OPENGATEWAY_API_KEY",
 	venice: "VENICE_API_KEY",
 	vllm: "VLLM_API_KEY",
 	xiaomi: "XIAOMI_API_KEY",

package/src/types.ts

@@ -147,6 +147,7 @@ export type KnownProvider =
 	| "minimax"
 	| "opencode-go"
 	| "opencode-zen"
+	| "opengateway"
 	| "synthetic"
 	| "cloudflare-ai-gateway"
 	| "huggingface"

package/src/utils/idle-iterator.ts

@@ -2,6 +2,11 @@ import { $env } from "@gajae-code/utils";
 
 const DEFAULT_STREAM_IDLE_TIMEOUT_MS = 120_000;
 const DEFAULT_STREAM_FIRST_EVENT_TIMEOUT_MS = 100_000;
+const KIMI_CODE_FIRST_EVENT_TIMEOUT_MS = 300_000;
+
+export function getProviderFirstEventTimeoutFallbackMs(provider: string): number | undefined {
+	return provider === "kimi-code" ? KIMI_CODE_FIRST_EVENT_TIMEOUT_MS : undefined;
+}
 
 function normalizeIdleTimeoutMs(value: string | undefined, fallback: number): number | undefined {
 	if (value === undefined) return fallback;

package/src/utils/oauth/index.ts

@@ -240,6 +240,11 @@ const builtInOAuthProviders: OAuthProviderInfo[] = [
 		name: "ZenMux",
 		available: true,
 	},
+	{
+		id: "opengateway",
+		name: "OpenGateway by Sionic AI",
+		available: true,
+	},
 	{
 		id: "vllm",
 		name: "vLLM (Local OpenAI-compatible)",
@@ -386,6 +391,7 @@ export async function refreshOAuthToken(
 		case "vercel-ai-gateway":
 		case "qwen-portal":
 		case "zenmux":
+		case "opengateway":
 		case "vllm":
 			// API keys / static bearer tokens don't expire, return as-is
 			newCredentials = credentials;

package/src/utils/oauth/opengateway.ts

@@ -0,0 +1,15 @@
+/** OpenGateway (by Sionic AI) login flow (API key paste, validated via /v1/models). */
+import { createApiKeyLogin } from "./api-key-login";
+
+export const loginOpenGateway = createApiKeyLogin({
+	providerLabel: "OpenGateway by Sionic AI",
+	authUrl: "https://opengateway.ai/dashboard",
+	instructions: "Create or copy your OpenGateway API key",
+	promptMessage: "Paste your OpenGateway API key",
+	placeholder: "sk-...",
+	validation: {
+		kind: "models-endpoint",
+		provider: "OpenGateway by Sionic AI",
+		modelsUrl: "https://apis.opengateway.ai/v1/models",
+	},
+});

package/src/utils/oauth/types.ts

@@ -40,6 +40,7 @@ export type OAuthProvider =
 	| "openai-codex-device"
 	| "opencode-go"
 	| "opencode-zen"
+	| "opengateway"
 	| "parallel"
 	| "perplexity"
 	| "qianfan"