cartethyia 2.6.7 → 2.7.1

package/src/providers/integrations/buddy/workbuddy/shared.ts

@@ -8,12 +8,18 @@ import type { ProviderDispatchContext, ProviderDispatchTarget, ProviderId } from
 import { resolveInboundSessionId } from "../../../operations/session-resolution";
 import { buddyConversationHeaders } from "../buddy-conversation-headers";
 import { buddyCliHeaders, buddyCliTraceHeaders } from "../buddy-cli-headers";
-import { buddyEnterprise, buddyUserId, resolveBuddyIdentity } from "../buddy-identity";
+import { buddyEnterprise, buddyUserId, resolveBuddyDeviceBinding, resolveBuddyIdentity } from "../buddy-identity";
 import {
   buildWorkBuddyUserAgent,
   getWorkBuddyClientVersion,
   getWorkBuddyCliVersion,
 } from "../../../operations/client-versions";
+import {
+  getTuringDeviceTokenService,
+  resolveTuringSdkRoot,
+  type TuringSdkConfig,
+  WORKBUDDY_TURING_CONFIG,
+} from "../turing-device-token";
 
 /**
  * WorkBuddy serves chat from a versioned path on a base URL that carries no
@@ -140,8 +146,17 @@ export async function workbuddyHeaders(
     headers["x-machine-id"] = deriveAccountStableId(accountId, "machine");
     headers["x-session-id"] = deriveAccountStableId(accountId, "session");
   }
-  const deviceToken = readBuddyDeviceToken(context);
-  if (deviceToken !== undefined) headers["x-device-token"] = deviceToken;
+  // Per-account device binding: distinct device identity per account (see
+  // resolveBuddyDeviceBinding). Sent as the desktop's X-Device-Id-List.
+  const binding = resolveBuddyDeviceBinding(context);
+  if (binding !== undefined) {
+    headers["x-device-id-list"] = JSON.stringify({ qimei36: binding });
+  }
+  Object.assign(headers, await buddyDeviceTokenHeaders({
+    realm: domain.endsWith(".cn") ? "cn" : "intl",
+    context,
+    cliIdentity,
+  }));
   return headers;
 }
 
@@ -165,6 +180,44 @@ export function readBuddyDeviceToken(context?: ProviderDispatchContext): string
   return readDeviceTokenFile();
 }
 
+/**
+ * The reference `X-Device-Token` pair, per the official desktop app.
+ *
+ * The desktop client (WorkBuddyAI, channel 400111) always presents one of the
+ * two on chat traffic — a real token from the TuringShield SDK, or an
+ * `X-Device-Token-Error` when the SDK cannot produce one. It never omits both,
+ * and it sends the header on every realm (Intl and CN alike). The SDK is a
+ * proprietary native addon; when it is not configured we fall back to a
+ * captured token (auth_state / `CARTETHYIA_DEVICE_TOKEN_FILE`) and only then
+ * to the official failure state.
+ *
+ * - `CARTETHYIA_TURING_SDK_PATH` → real token via `TuringDeviceTokenService`.
+ * - Legacy token (auth_state/file) → used when the SDK is unavailable.
+ * - CLI identity never carries a device token (the CLI has no device).
+ */
+export async function buddyDeviceTokenHeaders(args: {
+  realm: "intl" | "cn";
+  context: ProviderDispatchContext | undefined;
+  cliIdentity: boolean | undefined;
+  turingConfig?: TuringSdkConfig;
+}): Promise<Record<string, string>> {
+  const { context, cliIdentity, turingConfig } = args;
+  void args.realm; // the desktop client sends the token on both realms
+  if (resolveBuddyIdentity(context, cliIdentity) === "cli") return {};
+  // Native SDK first (real token), when configured.
+  if (resolveTuringSdkRoot() !== null) {
+    const sdkHeaders = await getTuringDeviceTokenService(turingConfig ?? WORKBUDDY_TURING_CONFIG).headersForRequest(false);
+    if (sdkHeaders["x-device-token"] !== undefined) return sdkHeaders;
+    // SDK configured but could not produce a token: mirror its failure state,
+    // like the desktop client does — do not fall back to a stale file token.
+    return sdkHeaders;
+  }
+  // No SDK on this deployment: a captured token is better than nothing.
+  const token = readBuddyDeviceToken(context);
+  if (token !== undefined) return { "x-device-token": token };
+  return { "x-device-token-error": "sdk_not_initialized" };
+}
+
 function readDeviceTokenFile(): string | undefined {
   const now = Date.now();
   if (deviceTokenFileCache !== undefined && now - deviceTokenFileCache.readAt < DEVICE_TOKEN_FILE_TTL_MS) {

package/src/providers/integrations/claude/claude.ts

@@ -49,7 +49,7 @@ import { getClaudeCliVersion, adoptClaudeCliVersion, getClaudeSdkVersion } from
 import { adoptRequiredVersion } from "../../operations/client-version-resolver";
 import { getOrCreateInstallId } from "../shared/install-id";
 import { canonicalToClaudeMessagesPayload } from "../../../protocol/request/messages";
-import { applyParamQuirks } from "../../../transport/translation/quirks";
+import { applyParamQuirks } from "../../../protocol/translation/quirks";
 import { CLAUDE_BILLING_HEADER_PREFIX, endpointUrl } from "../../../protocol/codecs";
 import { sendClaudeMessagesRequest } from "../../../protocol/upstream/send-claude-messages";
 import {

package/src/providers/integrations/cloudflareai/cloudflareai.ts

@@ -25,7 +25,7 @@ import type { CanonicalEvent, CanonicalRequest } from "../../../protocol/canonic
 import { GatewayError } from "../../../protocol/gateway-error";
 import { buildClaudeMessagesRequest, filterClaudeCustomHeaders } from "../shared/claude-messages";
 import { sendClaudeMessagesRequest } from "../../../protocol/upstream/send-claude-messages";
-import { applyParamQuirks } from "../../../transport/translation/quirks";
+import { applyParamQuirks } from "../../../protocol/translation/quirks";
 import { decodeCloudflareAiGatewayCredential } from "../../../security/credential-codec";
 
 export const CLOUDFLAREAI_PROVIDER_ID = "cloudflareai" as const;

package/src/providers/integrations/codex/codex-errors.ts

@@ -7,11 +7,11 @@
 import { GatewayError } from "../../../protocol/gateway-error";
 import {
   classifyUpstreamError,
-  parseRetryAfter,
   extractUpstreamMessage,
   upstreamErrorIdentifier,
   upstreamRequestId,
 } from "../../../transport/failure-policy";
+import { parseRetryAfter } from "../../../protocol/upstream-parsing";
 import { tryParseJsonObject } from "../../../protocol/codecs";
 
 function toNumberOrUndefined(value: string | null): number | undefined {

package/src/providers/integrations/codex/codex.ts

@@ -33,9 +33,9 @@ import { isRecord, joinUrl } from "../../../protocol/codecs";
 import { postUpstreamJson } from "../../../protocol/upstream/post-json";
 import {
   MIN_RETRY_BUDGET_MS,
-  parseUpstreamBackoff,
   sleep as defaultSleep,
 } from "../../../transport/failure-policy";
+import { parseUpstreamBackoff } from "../../../protocol/upstream-parsing";
 
 /**
  * Sleep used by the 429 capacity-wait loop, overridable in tests.

package/src/providers/integrations/commandcode/commandcode.ts

@@ -1,5 +1,6 @@
 // Factory-blocked (Phase C4): proprietary `{threadId, config, params}`
 // envelope with NDJSON stream transform — non-OpenAI wire; stays bespoke.
+import { randomUUID } from "node:crypto";
 import { GatewayError } from "../../../protocol/gateway-error";
 import { canContainToolResult, joinTextParts, toolCallParts, toolResultParts } from "../../../protocol/canonical-model";
 import {
@@ -21,6 +22,21 @@ import {
 import { MimoThinkSplitter } from "../xiaomi-mimo/think-stream";
 export const COMMANDCODE_PROVIDER_ID = "commandcode" as const;
 export const COMMANDCODE_BASE_URL = providerBaseUrl("commandcode");
+/**
+ * The host alone. `/provider/v1` (the catalog base) is a sibling of the
+ * generate surface, so dispatch resolves paths against the host instead:
+ * appending `/alpha/generate` to `/provider/v1` produced a URL that is
+ * neither surface.
+ */
+export const COMMANDCODE_HOST = "https://api.commandcode.ai" as const;
+/**
+ * The CLI's generate surface, and the one the reference client calls for every
+ * model — it does not vary the path by the wire family a model row declares.
+ * A completion dispatched to `/provider/v1/<wire>` was reaching the catalog
+ * surface, which serves `GET /models` and is not what the CLI calls to
+ * generate.
+ */
+export const COMMANDCODE_GENERATE_PATH = "/alpha/generate" as const;
 /** The three routes the Provider API serves, one per wire family. */
 export const COMMANDCODE_CHAT_PATH = "/chat/completions" as const;
 export const COMMANDCODE_RESPONSES_PATH = "/responses" as const;
@@ -43,101 +59,101 @@ export const COMMANDCODE_MESSAGES_PATH = "/messages" as const;
  * rather than lingering as a card that can only ever fail.
  */
 export const COMMANDCODE_MODELS: readonly ModelDefinition[] = [
-  defineModel({ id: "MiniMaxAI/MiniMax-M2.5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 200000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "MiniMaxAI/MiniMax-M2.7", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 200000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "MiniMaxAI/MiniMax-M3", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "Qwen/Qwen3.6-Max-Preview", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 200000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "Qwen/Qwen3.6-Plus", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 200000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "Qwen/Qwen3.7-Flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "Qwen/Qwen3.7-Max", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "Qwen/Qwen3.7-Plus", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "Qwen/Qwen3.8-27B", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 262144, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "Qwen/Qwen3.8-Flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: "/chat/completions", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "Qwen/Qwen3.8-Max", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "Qwen/Qwen3.8-Max-0902", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: "/chat/completions", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "Qwen/Qwen3.8-Max-Prime", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "Qwen/Qwen3.8-Omni-Flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "claude-fable-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: "/messages", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "claude-fable-5-1", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: "/messages", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "claude-haiku-4-5-20251001", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: "/messages", ctx: 200000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "claude-haiku-5-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: "/messages", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "claude-opus-4-7", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: "/messages", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "claude-opus-4-8", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: "/messages", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "claude-opus-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: "/messages", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "claude-opus-5-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: "/messages", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "claude-sonnet-4-6", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: "/messages", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "claude-sonnet-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: "/messages", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "claude-sonnet-5-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: "/messages", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "deepseek/deepseek-v4-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "deepseek/deepseek-v4-flash-fast", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: "/chat/completions", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "deepseek/deepseek-v4-flash-vision-exp", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "deepseek/deepseek-v4-pro", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "deepseek/deepseek-v4.1-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "deepseek/deepseek-v4.1-flash-fast", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "google/gemini-3.1-flash-lite", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "google/gemini-3.5-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "google/gemini-3.5-flash-lite", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "google/gemini-3.6-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "google/gemini-3.7-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: "/chat/completions", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "google/gemini-3.8-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "google/gemini-4-argon", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "gpt-5.3-codex", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 400000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "gpt-5.4", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 400000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "gpt-5.4-mini", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 400000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "gpt-5.5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 400000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "gpt-5.6-luna", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "gpt-5.6-sol", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "gpt-5.6-terra", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "gpt-6-astra", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "gpt-6-luna", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "gpt-6-sol", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "gpt-6.1-sol", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "inclusionai/ling-3.1-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: "/chat/completions", ctx: 262144, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "meituan/LongCat-2.0", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: "/chat/completions", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "meta/muse-spark-1.1", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "meta/muse-spark-1.2", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "meta/muse-spark-1.2-contributor", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "meta/muse-spark-1.3", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "meta/muse-spark-1.3-contributor", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "meta/muse-spark-open", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "meta/muse-wonder-1.0", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "minimax/minimax-m3.1-flash-preview", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "minimax/minimax-m3.1-flash-preview:free", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "mistral/mistral-large-4", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 524288, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "moonshotai/Kimi-K2.5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 256000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "moonshotai/Kimi-K2.6", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 256000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "moonshotai/Kimi-K2.7-Code", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 256000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "moonshotai/Kimi-K2.7-Code-Highspeed", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 262000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "moonshotai/Kimi-K3", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "nvidia/nemotron-3-ultra-550b-a55b", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "nvidia/nemotron-3.5-super-vl", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 262144, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "poolside/laguna-s-2.1-free", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 256000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "reflection/beam:free", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "sakana/fugu-ultra", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "stealth/cat-jam:free", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "stealth/glyph-cluster:free", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 256000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "stepfun/Step-3.5-Flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 262144, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "stepfun/Step-3.7-Flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 256000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "stepfun/Step-5-Preview", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "tencent/hy3-paid", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 262144, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "tencent/hy4-preview", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: "/chat/completions", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "thinkingmachines/inkling", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 256000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "thinkingmachines/inkling-small", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "xai/grok-4.5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 500000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "xai/grok-4.6", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 500000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "xai/grok-4.7", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 500000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "xiaomi/mimo-v2.5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "xiaomi/mimo-v2.5-pro", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "xiaomi/mimo-v2.6-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "xiaomi/mimo-v2.6-pro", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "xiaomi/mimo-v2.6-pro-ultraspeed", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "z-ai/glm-5.3-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: "/chat/completions", ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "z-ai/glm-5.3-flashx", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "zai-org/GLM-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 200000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "zai-org/GLM-5.1", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 200000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "zai-org/GLM-5.2", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "zai-org/GLM-5.2-Fast", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
-  defineModel({ id: "zai-org/GLM-5.3", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: "/responses", ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "MiniMaxAI/MiniMax-M2.5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 200000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "MiniMaxAI/MiniMax-M2.7", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 200000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "MiniMaxAI/MiniMax-M3", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "Qwen/Qwen3.6-Max-Preview", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 200000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "Qwen/Qwen3.6-Plus", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 200000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "Qwen/Qwen3.7-Flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "Qwen/Qwen3.7-Max", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "Qwen/Qwen3.7-Plus", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "Qwen/Qwen3.8-27B", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 262144, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "Qwen/Qwen3.8-Flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "Qwen/Qwen3.8-Max", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "Qwen/Qwen3.8-Max-0902", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "Qwen/Qwen3.8-Max-Prime", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "Qwen/Qwen3.8-Omni-Flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "claude-fable-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "claude-fable-5-1", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "claude-haiku-4-5-20251001", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 200000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "claude-haiku-5-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "claude-opus-4-7", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "claude-opus-4-8", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "claude-opus-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "claude-opus-5-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "claude-sonnet-4-6", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "claude-sonnet-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "claude-sonnet-5-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "messages", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "deepseek/deepseek-v4-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "deepseek/deepseek-v4-flash-fast", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "deepseek/deepseek-v4-flash-vision-exp", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "deepseek/deepseek-v4-pro", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "deepseek/deepseek-v4.1-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "deepseek/deepseek-v4.1-flash-fast", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "google/gemini-3.1-flash-lite", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "google/gemini-3.5-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "google/gemini-3.5-flash-lite", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "google/gemini-3.6-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "google/gemini-3.7-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "google/gemini-3.8-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "google/gemini-4-argon", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "gpt-5.3-codex", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 400000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "gpt-5.4", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 400000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "gpt-5.4-mini", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 400000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "gpt-5.5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 400000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "gpt-5.6-luna", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "gpt-5.6-sol", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "gpt-5.6-terra", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "gpt-6-astra", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "gpt-6-luna", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "gpt-6-sol", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "gpt-6.1-sol", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1050000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "inclusionai/ling-3.1-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 262144, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "meituan/LongCat-2.0", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "meta/muse-spark-1.1", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "meta/muse-spark-1.2", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "meta/muse-spark-1.2-contributor", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "meta/muse-spark-1.3", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "meta/muse-spark-1.3-contributor", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "meta/muse-spark-open", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "meta/muse-wonder-1.0", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "minimax/minimax-m3.1-flash-preview", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "minimax/minimax-m3.1-flash-preview:free", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "mistral/mistral-large-4", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 524288, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "moonshotai/Kimi-K2.5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 256000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "moonshotai/Kimi-K2.6", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 256000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "moonshotai/Kimi-K2.7-Code", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 256000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "moonshotai/Kimi-K2.7-Code-Highspeed", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 262000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "moonshotai/Kimi-K3", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "nvidia/nemotron-3-ultra-550b-a55b", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "nvidia/nemotron-3.5-super-vl", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 262144, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "poolside/laguna-s-2.1-free", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 256000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "reflection/beam:free", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "sakana/fugu-ultra", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "stealth/cat-jam:free", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "stealth/glyph-cluster:free", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 256000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "stepfun/Step-3.5-Flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 262144, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "stepfun/Step-3.7-Flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 256000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "stepfun/Step-5-Preview", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "tencent/hy3-paid", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 262144, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "tencent/hy4-preview", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "thinkingmachines/inkling", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 256000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "thinkingmachines/inkling-small", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "xai/grok-4.5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 500000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "xai/grok-4.6", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 500000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "xai/grok-4.7", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 500000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "xiaomi/mimo-v2.5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "xiaomi/mimo-v2.5-pro", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "xiaomi/mimo-v2.6-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "xiaomi/mimo-v2.6-pro", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "xiaomi/mimo-v2.6-pro-ultraspeed", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "z-ai/glm-5.3-flash", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "chat", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1048576, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "z-ai/glm-5.3-flashx", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "zai-org/GLM-5", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 200000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "zai-org/GLM-5.1", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 200000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "zai-org/GLM-5.2", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "zai-org/GLM-5.2-Fast", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
+  defineModel({ id: "zai-org/GLM-5.3", providerId: COMMANDCODE_PROVIDER_ID, wireFamily: "responses", endpoint: COMMANDCODE_GENERATE_PATH, ctx: 1000000, reasoning: true, toolCall: true, vision: true }),
 ];
 
 
@@ -615,11 +631,13 @@ export async function* transformNdjson(body: ReadableStream<Uint8Array>, signal:
 // ProviderAdapter — the Provider API serves three wires, so dispatch is the
 // shared OpenAI-compatible adapter rather than one bespoke NDJSON reader.
 //
-// `/provider/v1` answers `/chat/completions`, `/responses` and `/messages`, and
-// the model row states which one serves it. The bespoke `/alpha/generate`
-// envelope is the CLI's own surface and stays reachable through
-// `buildRequest` for callers that still speak it, while routing follows the
-// wire each model row declares rather than forcing every model onto it.
+// Generation goes to `/alpha/generate` for every model: that is the CLI's own
+// surface and the one the reference client calls, and it does not vary the path
+// by the wire a model row declares. `/provider/v1` is the catalog surface — it
+// answers `GET /models`, which discovery still uses — so a completion posted to
+// `/provider/v1/<wire>` was reaching a surface the CLI never calls to
+// generate. The wire family still decides how the body is shaped; only the
+// endpoint is shared.
 
 interface CommandCodeAdapterOptions {
   readonly fetch?: typeof fetch;
@@ -632,6 +650,11 @@ function commandCodeIdentityHeaders(_token: string): Record<string, string> {
     "x-taste-learning": "true",
     "x-project-slug": "cartethyia",
     "user-agent": "cli",
+    // Per-request, as the CLI client sends it. The upstream expects a session
+    // id on every generate call and the reference client mints a fresh UUID
+    // for each one; sending a constant correlated requests that had nothing
+    // to do with each other.
+    "x-session-id": randomUUID(),
   };
 }
 
@@ -639,11 +662,18 @@ function providerApiAdapter(options: CommandCodeAdapterOptions): ProviderAdapter
   return createApiKeyAdapter(
     {
       provider_id: COMMANDCODE_PROVIDER_ID,
-      base_url: COMMANDCODE_BASE_URL,
+      // The host, not the `/provider/v1` base: generation is a sibling
+      // surface, so every wire resolves under the host and the path below
+      // supplies the CLI endpoint. Nesting it under `/provider/v1` produced a
+      // URL that is neither surface.
+      base_url: COMMANDCODE_HOST,
       endpoint_paths_by_wire_family: {
-        chat: COMMANDCODE_CHAT_PATH,
-        responses: COMMANDCODE_RESPONSES_PATH,
-        messages: COMMANDCODE_MESSAGES_PATH,
+        // Every wire, not one per family: the reference client sends all
+        // generation to `/alpha/generate` regardless of the row's declared
+        // wire, so dispatch must not fan out across three paths.
+        chat: COMMANDCODE_GENERATE_PATH,
+        responses: COMMANDCODE_GENERATE_PATH,
+        messages: COMMANDCODE_GENERATE_PATH,
       },
       buildExtraHeaders: async (context: ProviderDispatchContext) =>
         commandCodeIdentityHeaders(await readCredentialSecret(context.credential, "Command Code bearer token is required")),

package/src/providers/integrations/commandcode/discovery.ts

@@ -30,14 +30,13 @@ import { isRecord } from "../../../protocol/codecs";
 import { boundedUpstreamNumber, sanitizeUpstreamLabel } from "../../provider-sanitize";
 import type { ProviderModelDiscovery } from "../../discovery/discovery-types";
 import { defineModel } from "../../provider-model-definition";
-import { COMMANDCODE_PROVIDER_ID } from "./commandcode";
+import { COMMANDCODE_GENERATE_PATH, COMMANDCODE_PROVIDER_ID } from "./commandcode";
 
 /** The Provider API's OpenAI-shaped model listing. */
 export const COMMANDCODE_MODELS_PATH = "/models" as const;
 
 /** The route that selects the OpenAI Responses wire when a model lists it. */
 const RESPONSES_ENDPOINT = "/responses" as const;
-const ANTHROPIC_MESSAGES_ENDPOINT = "/messages" as const;
 
 const MAX_CONTEXT = 1_000_000_000;
 
@@ -54,12 +53,14 @@ export function commandCodeWireFor(
   supportedEndpoints?: readonly string[],
 ): { readonly wireFamily: "messages" | "responses" | "chat"; readonly endpointPath: string } {
   if (modelId.startsWith("claude-")) {
-    return { wireFamily: "messages", endpointPath: ANTHROPIC_MESSAGES_ENDPOINT };
+    // The wire family still decides how the body is shaped; only the endpoint
+    // is the CLI's, for every row. See `COMMANDCODE_GENERATE_PATH`.
+    return { wireFamily: "messages", endpointPath: COMMANDCODE_GENERATE_PATH };
   }
   if (supportedEndpoints?.includes(RESPONSES_ENDPOINT)) {
-    return { wireFamily: "responses", endpointPath: RESPONSES_ENDPOINT };
+    return { wireFamily: "responses", endpointPath: COMMANDCODE_GENERATE_PATH };
   }
-  return { wireFamily: "chat", endpointPath: "/chat/completions" };
+  return { wireFamily: "chat", endpointPath: COMMANDCODE_GENERATE_PATH };
 }
 
 /** Reads `supported_endpoints` tolerantly — unknown rows keep the chat wire. */

package/src/providers/integrations/factory/factory.ts

@@ -49,7 +49,7 @@ import { createUpstreamDeadlineLifecycle, abortGatewayError } from "../../operat
 import { mapUpstreamHttpError } from "../../../transport/failure-policy";
 import { buildClaudeMessagesRequest, filterClaudeCustomHeaders } from "../shared/claude-messages";
 import { sendClaudeMessagesRequest } from "../../../protocol/upstream/send-claude-messages";
-import { applyParamQuirks } from "../../../transport/translation/quirks";
+import { applyParamQuirks } from "../../../protocol/translation/quirks";
 import { createApiKeyAdapter } from "../../configured-provider";
 import type { ApiKeyProviderSpec } from "../../configured-provider";
 import { decodeFactoryDroidCredential } from "../../../security/credential-codec";

package/src/providers/integrations/github/github.ts

@@ -25,7 +25,7 @@ import { postUpstreamJson } from "../../../protocol/upstream/post-json";
 import { providerBaseUrl } from "../../provider-metadata";
 import { getOrCreateInstallId } from "../shared/install-id";
 import { resolveRedisClient } from "../../../persistence/redis";
-import { applyParamQuirks } from "../../../transport/translation/quirks";
+import { applyParamQuirks } from "../../../protocol/translation/quirks";
 import { resolveInboundSessionId } from "../../operations/session-resolution";
 import {
   GITHUB_REQUEST_HEADERS,

package/src/providers/integrations/grok/grok.ts

@@ -20,7 +20,7 @@ import { providerBaseUrl } from "../../provider-metadata";
 import { boundedUpstreamArray, boundedUpstreamNumber, sanitizeUpstreamLabel } from "../../provider-sanitize";
 import { isRecord } from "../../../protocol/codecs";
 import { modelsDevCatalog } from "../../discovery/models-dev-catalog";
-import type { ReasoningEffortLevel } from "../../../transport/translation/thinking";
+import type { ReasoningEffortLevel } from "../../../protocol/translation/thinking";
 import type { FetchLike } from "../../quota/quota-contracts";
 import { resolveInboundSessionId, resolvePromptCacheKey } from "../../operations/session-resolution";
 import {