From 6563d6d2a53b112088cb04947f8b2df03bedb1ae Mon Sep 17 00:00:00 2001
From: Michael Suchacz <203725896+ibetitsmike@users.noreply.github.com>
Date: Fri, 17 Jul 2026 17:27:19 +0000
Subject: [PATCH 1/4] feat: add native support for Kimi K3 via OpenRouter

---
 docs/config/models.mdx                        |  1 +
 src/common/constants/knownModels.test.ts      | 14 ++++++++--
 src/common/constants/knownModels.ts           | 13 ++++++++-
 src/common/types/thinking.ts                  |  9 ++++++
 src/common/utils/ai/modelCapabilities.test.ts |  7 +++++
 src/common/utils/ai/providerOptions.test.ts   | 28 +++++++++++++++++++
 src/common/utils/ai/providerOptions.ts        | 11 ++++++--
 src/common/utils/thinking/policy.test.ts      | 11 ++++++++
 src/common/utils/thinking/policy.ts           |  7 +++++
 src/common/utils/tokens/modelStats.test.ts    |  9 ++++++
 src/common/utils/tokens/models-extra.ts       | 16 +++++++++++
 .../builtInSkillContent.generated.ts          |  1 +
 12 files changed, 120 insertions(+), 7 deletions(-)

diff --git a/docs/config/models.mdx b/docs/config/models.mdx
index 8bc9407d6e..06c2dd1ba5 100644
--- a/docs/config/models.mdx
+++ b/docs/config/models.mdx
@@ -34,6 +34,7 @@ Mux ships with curated models kept up to date with the frontier. Use any custom
 | Grok Code Fast 1       | xai:grok-code-fast-1          | `grok-code`                                                  |         |
 | DeepSeek V4 Pro        | deepseek:deepseek-v4-pro      | `deepseek`, `deepseek-pro`, `deepseek-v4`, `deepseek-v4-pro` |         |
 | DeepSeek V4 Flash      | deepseek:deepseek-v4-flash    | `deepseek-flash`, `deepseek-v4-flash`                        |         |
+| Kimi K3                | openrouter:moonshotai/kimi-k3 | `kimi`, `k3`, `kimi-k3`                                      |         |
 
 {/* END KNOWN_MODELS_TABLE */}
 
diff --git a/src/common/constants/knownModels.test.ts b/src/common/constants/knownModels.test.ts
index a753e45077..092a0b3725 100644
--- a/src/common/constants/knownModels.test.ts
+++ b/src/common/constants/knownModels.test.ts
@@ -14,9 +14,12 @@ describe("Known Models Integration", () => {
     for (const [key, model] of Object.entries(KNOWN_MODELS)) {
       const modelId = model.providerModelId;
 
-      // xAI models are prefixed with "xai/" in models.json.
-      const lookupKey = model.provider === "xai" ? `xai/${modelId}` : modelId;
-      if (!(lookupKey in modelsJson) && !(modelId in modelsExtra)) {
+      // xAI and OpenRouter models are provider-prefixed in token metadata.
+      const lookupKey =
+        model.provider === "xai" || model.provider === "openrouter"
+          ? `${model.provider}/${modelId}`
+          : modelId;
+      if (!(lookupKey in modelsJson) && !(lookupKey in modelsExtra) && !(modelId in modelsExtra)) {
         missingModels.push(`${key}: ${model.provider}:${modelId}`);
       }
     }
@@ -43,6 +46,11 @@ describe("Known Models Integration", () => {
     expect(MODEL_ABBREVIATIONS["gpt-5.5"]).toBeUndefined();
   });
 
+  test("kimi aliases resolve to the OpenRouter-scoped Kimi K3 model", () => {
+    expect(MODEL_ABBREVIATIONS.kimi).toBe("openrouter:moonshotai/kimi-k3");
+    expect(MODEL_ABBREVIATIONS.k3).toBe("openrouter:moonshotai/kimi-k3");
+  });
+
   test("known model ids and aliases stay unique across the curated registry", () => {
     const seenIds = new Set<string>();
     const seenAliases = new Set<string>();
diff --git a/src/common/constants/knownModels.ts b/src/common/constants/knownModels.ts
index 3d0029fa35..1542a311cf 100644
--- a/src/common/constants/knownModels.ts
+++ b/src/common/constants/knownModels.ts
@@ -4,7 +4,7 @@
 
 import { formatModelDisplayName } from "../utils/ai/modelDisplay";
 
-type ModelProvider = "anthropic" | "openai" | "google" | "xai" | "deepseek";
+type ModelProvider = "anthropic" | "openai" | "google" | "xai" | "deepseek" | "openrouter";
 
 interface KnownModelDefinition {
   /** Provider identifier used by SDK factories */
@@ -211,6 +211,17 @@ const MODEL_DEFINITIONS = {
     aliases: ["deepseek-flash", "deepseek-v4-flash"],
     tokenizerOverride: "deepseek/deepseek-v3.1",
   },
+  // Kimi K3 - Moonshot AI's flagship open-weight multimodal reasoning model (released
+  // July 16, 2026; 1M context, text+image input; $3/M in, $15/M out via OpenRouter).
+  // Bare `kimi` alias tracks Moonshot's flagship per the shortest-alias convention.
+  KIMI_K3: {
+    provider: "openrouter",
+    providerModelId: "moonshotai/kimi-k3",
+    aliases: ["kimi", "k3", "kimi-k3"],
+    // K3's tokenizer isn't published in ai-tokenizer yet; reuse Kimi K2 (the newest
+    // Moonshot tokenizer available) for approximate counting.
+    tokenizerOverride: "moonshotai/kimi-k2",
+  },
 } as const satisfies Record<string, KnownModelDefinition>;
 
 export type KnownModelKey = keyof typeof MODEL_DEFINITIONS;
diff --git a/src/common/types/thinking.ts b/src/common/types/thinking.ts
index f8139247c4..09d8b57632 100644
--- a/src/common/types/thinking.ts
+++ b/src/common/types/thinking.ts
@@ -304,6 +304,15 @@ export function openaiSupportsProMode(modelString: string): boolean {
   return isGpt56FamilyModel(modelString);
 }
 
+/**
+ * Kimi K3 (Moonshot AI via OpenRouter) always reasons, and OpenRouter currently
+ * exposes only its max (default) reasoning effort; the thinking policy and the
+ * OpenRouter provider-options branch both key off this predicate.
+ */
+export function isKimiK3Model(modelString: string): boolean {
+  return stripModelProviderPrefixes(modelString) === "kimi-k3";
+}
+
 /**
  * Whether the given Anthropic model rejects `thinking: { type: "disabled" }`.
  *
diff --git a/src/common/utils/ai/modelCapabilities.test.ts b/src/common/utils/ai/modelCapabilities.test.ts
index ff06345327..d5492e9585 100644
--- a/src/common/utils/ai/modelCapabilities.test.ts
+++ b/src/common/utils/ai/modelCapabilities.test.ts
@@ -33,6 +33,13 @@ describe("getModelCapabilities", () => {
     expect(caps).not.toBeNull();
   });
 
+  it("reports image support for Kimi K3 via the OpenRouter gateway id", () => {
+    const caps = getModelCapabilities("openrouter:moonshotai/kimi-k3");
+    expect(caps).not.toBeNull();
+    expect(caps?.supportsVision).toBe(true);
+    expect(caps?.supportsPdfInput).toBe(false);
+  });
+
   it("infers PDF support for OpenAI vision models when models-extra omits the flag", () => {
     const caps = getModelCapabilities("openai:gpt-5.5");
     expect(caps).not.toBeNull();
diff --git a/src/common/utils/ai/providerOptions.test.ts b/src/common/utils/ai/providerOptions.test.ts
index 64cb6ffdf4..dcf9eba0de 100644
--- a/src/common/utils/ai/providerOptions.test.ts
+++ b/src/common/utils/ai/providerOptions.test.ts
@@ -1564,6 +1564,34 @@ describe("buildProviderOptions - Google", () => {
   });
 });
 
+describe("buildProviderOptions - OpenRouter", () => {
+  test("enables reasoning without an effort override for Kimi K3", () => {
+    // K3 accepts only its default (max) reasoning effort, so no effort value is sent.
+    expect(buildProviderOptions("openrouter:moonshotai/kimi-k3", "max")).toEqual({
+      openrouter: {
+        reasoning: {
+          enabled: true,
+          exclude: false,
+        },
+      },
+    });
+  });
+
+  test("maps thinking levels to effort for other OpenRouter reasoning models", () => {
+    expect(buildProviderOptions("openrouter:z-ai/glm-4.6", "medium")).toEqual({
+      openrouter: {
+        reasoning: {
+          enabled: true,
+          effort: "medium",
+          exclude: false,
+        },
+      },
+    });
+
+    expect(buildProviderOptions("openrouter:z-ai/glm-4.6", "off")).toEqual({});
+  });
+});
+
 describe("buildRequestHeaders", () => {
   for (const { name, model, options, expected } of [
     {
diff --git a/src/common/utils/ai/providerOptions.ts b/src/common/utils/ai/providerOptions.ts
index 77c7882037..b5cfa3e3f2 100644
--- a/src/common/utils/ai/providerOptions.ts
+++ b/src/common/utils/ai/providerOptions.ts
@@ -22,6 +22,7 @@ import {
   ANTHROPIC_THINKING_BUDGETS,
   GEMINI_THINKING_BUDGETS,
   getOpenAIReasoningEffort,
+  isKimiK3Model,
   openaiSupportsProMode,
   OPENROUTER_REASONING_EFFORT,
 } from "@/common/types/thinking";
@@ -474,7 +475,11 @@ export function buildProviderOptions(
 
   // Build OpenRouter-specific options
   if (formatProvider === "openrouter") {
-    const reasoningEffort = OPENROUTER_REASONING_EFFORT[effectiveThinking];
+    // Kimi K3 always reasons and OpenRouter currently accepts only its default (max)
+    // reasoning effort, so enable reasoning without an effort override instead of
+    // sending a low/medium/high value the model does not support.
+    const kimiK3 = isKimiK3Model(capabilityModel);
+    const reasoningEffort = kimiK3 ? undefined : OPENROUTER_REASONING_EFFORT[effectiveThinking];
 
     log.debug("buildProviderOptions: OpenRouter config", {
       reasoningEffort,
@@ -482,12 +487,12 @@ export function buildProviderOptions(
     });
 
     // Only add reasoning config if thinking is enabled
-    if (reasoningEffort) {
+    if (kimiK3 || reasoningEffort) {
       const options = {
         openrouter: {
           reasoning: {
             enabled: true,
-            effort: reasoningEffort,
+            ...(reasoningEffort ? { effort: reasoningEffort } : {}),
             // Don't exclude reasoning content - we want to display it in the UI
             exclude: false,
           },
diff --git a/src/common/utils/thinking/policy.test.ts b/src/common/utils/thinking/policy.test.ts
index 08abefeb3d..7f76f2d763 100644
--- a/src/common/utils/thinking/policy.test.ts
+++ b/src/common/utils/thinking/policy.test.ts
@@ -608,6 +608,17 @@ describe("getThinkingPolicyForModel", () => {
     ]);
   });
 
+  test("returns a fixed max policy for Kimi K3 behind OpenRouter", () => {
+    expect(getThinkingPolicyForModel("openrouter:moonshotai/kimi-k3")).toEqual(["max"]);
+    // Variant ids must not inherit the fixed K3 policy.
+    expect(getThinkingPolicyForModel("openrouter:moonshotai/kimi-k3-turbo")).toEqual([
+      "off",
+      "low",
+      "medium",
+      "high",
+    ]);
+  });
+
   test("returns off/low/medium/high for non-preview Gemini 3 Flash IDs", () => {
     for (const model of ["google:gemini-3-flash", "google:gemini-3-flash-001"]) {
       expect(getThinkingPolicyForModel(model)).toEqual(["off", "low", "medium", "high"]);
diff --git a/src/common/utils/thinking/policy.ts b/src/common/utils/thinking/policy.ts
index efe58b78ce..0fd2b2662a 100644
--- a/src/common/utils/thinking/policy.ts
+++ b/src/common/utils/thinking/policy.ts
@@ -19,6 +19,7 @@ import {
   THINKING_LEVEL_OFF,
   anthropicRejectsDisabledThinking,
   anthropicSupportsNativeXhigh,
+  isKimiK3Model,
   openaiSupportsNativeMaxEffort,
   stripModelProviderPrefixes,
   type ThinkingLevel,
@@ -168,6 +169,12 @@ function getExplicitThinkingPolicy(modelString: string): ThinkingPolicy | null {
     return ["low", "high"];
   }
 
+  // Kimi K3 always reasons and OpenRouter currently exposes only the max (default)
+  // reasoning effort, so the policy is a fixed single level.
+  if (isKimiK3Model(withoutProviderNamespace)) {
+    return ["max"];
+  }
+
   // No explicit reasoning rule matched.
   return null;
 }
diff --git a/src/common/utils/tokens/modelStats.test.ts b/src/common/utils/tokens/modelStats.test.ts
index c91ae74f95..de4103c64e 100644
--- a/src/common/utils/tokens/modelStats.test.ts
+++ b/src/common/utils/tokens/modelStats.test.ts
@@ -145,6 +145,15 @@ describe("getModelStats", () => {
     expect(flash.cache_read_input_token_cost).toBe(0.000000014);
   });
 
+  test("resolves Kimi K3 pricing and limits via the OpenRouter gateway id", () => {
+    const stats = expectStats("openrouter:moonshotai/kimi-k3");
+    expect(stats.max_input_tokens).toBe(1_048_576);
+    expect(stats.max_output_tokens).toBe(131_072);
+    expect(stats.input_cost_per_token).toBe(0.000003);
+    expect(stats.output_cost_per_token).toBe(0.000015);
+    expect(stats.cache_read_input_token_cost).toBe(0.0000003);
+  });
+
   test("resolves the default image generation model pricing", () => {
     const stats = expectStats(DEFAULT_IMAGE_MODEL);
 
diff --git a/src/common/utils/tokens/models-extra.ts b/src/common/utils/tokens/models-extra.ts
index 59b81b5b16..a54b8c440f 100644
--- a/src/common/utils/tokens/models-extra.ts
+++ b/src/common/utils/tokens/models-extra.ts
@@ -543,6 +543,22 @@ export const modelsExtra: Record<string, ModelData> = {
     supports_response_schema: true,
   },
 
+  // Kimi K3 via OpenRouter - released July 16, 2026. 1M context, 128K max output,
+  // text+image input (per the OpenRouter model page).
+  "openrouter/moonshotai/kimi-k3": {
+    max_input_tokens: 1048576,
+    max_output_tokens: 131072,
+    input_cost_per_token: 0.000003, // $3 per million input tokens
+    output_cost_per_token: 0.000015, // $15 per million output tokens
+    cache_read_input_token_cost: 0.0000003, // $0.30 per million cached input tokens
+    litellm_provider: "openrouter",
+    mode: "chat",
+    supports_function_calling: true,
+    supports_vision: true,
+    supports_reasoning: true,
+    supports_response_schema: true,
+  },
+
   // GPT-5.1-Codex-Max - Extended reasoning model with xhigh support
   // Same pricing as gpt-5.1-codex: $1.25/M input, $10/M output
   // Supports 5 reasoning levels: off, low, medium, high, xhigh
diff --git a/src/node/services/agentSkills/builtInSkillContent.generated.ts b/src/node/services/agentSkills/builtInSkillContent.generated.ts
index ee9937a53f..cd25b0be45 100644
--- a/src/node/services/agentSkills/builtInSkillContent.generated.ts
+++ b/src/node/services/agentSkills/builtInSkillContent.generated.ts
@@ -3188,6 +3188,7 @@ export const BUILTIN_SKILL_FILES: Record<string, Record<string, string>> = {
       "| Grok Code Fast 1       | xai:grok-code-fast-1          | `grok-code`                                                  |         |",
       "| DeepSeek V4 Pro        | deepseek:deepseek-v4-pro      | `deepseek`, `deepseek-pro`, `deepseek-v4`, `deepseek-v4-pro` |         |",
       "| DeepSeek V4 Flash      | deepseek:deepseek-v4-flash    | `deepseek-flash`, `deepseek-v4-flash`                        |         |",
+      "| Kimi K3                | openrouter:moonshotai/kimi-k3 | `kimi`, `k3`, `kimi-k3`                                      |         |",
       "",
       "{/* END KNOWN_MODELS_TABLE */}",
       "",

From 8e4a51ad6b813763aea9e9fc245598af791df94a Mon Sep 17 00:00:00 2001
From: Michael Suchacz <203725896+ibetitsmike@users.noreply.github.com>
Date: Fri, 17 Jul 2026 18:05:18 +0000
Subject: [PATCH 2/4] fix: send explicit max reasoning effort for Kimi K3
 (Codex review)

---
 src/common/types/thinking.ts                |  6 +++---
 src/common/utils/ai/providerOptions.test.ts |  6 ++++--
 src/common/utils/ai/providerOptions.ts      | 19 +++++++++++--------
 3 files changed, 18 insertions(+), 13 deletions(-)

diff --git a/src/common/types/thinking.ts b/src/common/types/thinking.ts
index 09d8b57632..79b0772e49 100644
--- a/src/common/types/thinking.ts
+++ b/src/common/types/thinking.ts
@@ -305,9 +305,9 @@ export function openaiSupportsProMode(modelString: string): boolean {
 }
 
 /**
- * Kimi K3 (Moonshot AI via OpenRouter) always reasons, and OpenRouter currently
- * exposes only its max (default) reasoning effort; the thinking policy and the
- * OpenRouter provider-options branch both key off this predicate.
+ * Kimi K3 (Moonshot AI via OpenRouter) always reasons and supports only the max
+ * reasoning effort; the thinking policy and the OpenRouter provider-options
+ * branch both key off this predicate.
  */
 export function isKimiK3Model(modelString: string): boolean {
   return stripModelProviderPrefixes(modelString) === "kimi-k3";
diff --git a/src/common/utils/ai/providerOptions.test.ts b/src/common/utils/ai/providerOptions.test.ts
index dcf9eba0de..9c4600667d 100644
--- a/src/common/utils/ai/providerOptions.test.ts
+++ b/src/common/utils/ai/providerOptions.test.ts
@@ -1565,12 +1565,14 @@ describe("buildProviderOptions - Google", () => {
 });
 
 describe("buildProviderOptions - OpenRouter", () => {
-  test("enables reasoning without an effort override for Kimi K3", () => {
-    // K3 accepts only its default (max) reasoning effort, so no effort value is sent.
+  test("sends the explicit max effort for Kimi K3", () => {
+    // `enabled: true` alone falls back to OpenRouter's default (medium) effort,
+    // which K3 does not support, so the effort must be sent explicitly.
     expect(buildProviderOptions("openrouter:moonshotai/kimi-k3", "max")).toEqual({
       openrouter: {
         reasoning: {
           enabled: true,
+          effort: "max",
           exclude: false,
         },
       },
diff --git a/src/common/utils/ai/providerOptions.ts b/src/common/utils/ai/providerOptions.ts
index b5cfa3e3f2..fb09d41b05 100644
--- a/src/common/utils/ai/providerOptions.ts
+++ b/src/common/utils/ai/providerOptions.ts
@@ -52,7 +52,9 @@ interface OpenRouterReasoningOptions {
   reasoning?: {
     enabled?: boolean;
     exclude?: boolean;
-    effort?: "low" | "medium" | "high";
+    // "max" is not in OpenRouter's generic low/medium/high set; it is the only
+    // effort Kimi K3 accepts (model metadata supported_efforts: ["max"]).
+    effort?: "low" | "medium" | "high" | "max";
   };
 }
 
@@ -475,11 +477,12 @@ export function buildProviderOptions(
 
   // Build OpenRouter-specific options
   if (formatProvider === "openrouter") {
-    // Kimi K3 always reasons and OpenRouter currently accepts only its default (max)
-    // reasoning effort, so enable reasoning without an effort override instead of
-    // sending a low/medium/high value the model does not support.
-    const kimiK3 = isKimiK3Model(capabilityModel);
-    const reasoningEffort = kimiK3 ? undefined : OPENROUTER_REASONING_EFFORT[effectiveThinking];
+    // Kimi K3 always reasons and supports only the max reasoning effort. Send it
+    // explicitly: `enabled: true` alone falls back to OpenRouter's default (medium)
+    // effort, which the model does not support.
+    const reasoningEffort = isKimiK3Model(capabilityModel)
+      ? "max"
+      : OPENROUTER_REASONING_EFFORT[effectiveThinking];
 
     log.debug("buildProviderOptions: OpenRouter config", {
       reasoningEffort,
@@ -487,12 +490,12 @@ export function buildProviderOptions(
     });
 
     // Only add reasoning config if thinking is enabled
-    if (kimiK3 || reasoningEffort) {
+    if (reasoningEffort) {
       const options = {
         openrouter: {
           reasoning: {
             enabled: true,
-            ...(reasoningEffort ? { effort: reasoningEffort } : {}),
+            effort: reasoningEffort,
             // Don't exclude reasoning content - we want to display it in the UI
             exclude: false,
           },

From 165664061c68f5931261e9d17e792fc1bb5e1966 Mon Sep 17 00:00:00 2001
From: Michael Suchacz <203725896+ibetitsmike@users.noreply.github.com>
Date: Fri, 17 Jul 2026 18:34:32 +0000
Subject: [PATCH 3/4] feat: switch Kimi K3 to a first-class Moonshot AI direct
 provider

---
 bun.lock                                      |  7 +++++
 docs/config/models.mdx                        |  2 +-
 docs/config/providers.mdx                     |  2 ++
 package.json                                  |  1 +
 scripts/gen_docs.ts                           |  1 +
 src/browser/assets/icons/moonshotai.svg       |  4 +++
 .../ProjectSidebar/ProjectSidebar.test.tsx    |  1 +
 .../components/ProviderIcon/ProviderIcon.tsx  |  2 ++
 .../Settings/Sections/ProvidersSection.tsx    |  1 +
 src/cli/run.ts                                |  2 +-
 src/common/config/schemas/providersConfig.ts  |  3 +++
 src/common/constants/knownModels.test.ts      | 10 +++----
 src/common/constants/knownModels.ts           |  8 +++---
 src/common/constants/providers.ts             | 10 ++++++-
 src/common/types/thinking.ts                  |  6 ++---
 src/common/utils/ai/modelCapabilities.test.ts | 12 +++++----
 src/common/utils/ai/providerOptions.test.ts   | 27 +++++++++++++++++--
 src/common/utils/ai/providerOptions.ts        | 24 +++++++++++++++++
 src/common/utils/thinking/policy.test.ts      |  5 ++--
 src/common/utils/thinking/policy.ts           |  4 +--
 src/common/utils/tokens/modelStats.test.ts    |  7 +++--
 src/common/utils/tokens/models-extra.ts       |  8 +++---
 .../builtInSkillContent.generated.ts          |  4 ++-
 src/node/utils/providerRequirements.ts        |  3 +++
 24 files changed, 121 insertions(+), 33 deletions(-)
 create mode 100644 src/browser/assets/icons/moonshotai.svg

diff --git a/bun.lock b/bun.lock
index 72f3cc5683..9306bc5cbb 100644
--- a/bun.lock
+++ b/bun.lock
@@ -11,6 +11,7 @@
         "@ai-sdk/deepseek": "^3.0.7",
         "@ai-sdk/google": "^4.0.11",
         "@ai-sdk/mcp": "^2.0.10",
+        "@ai-sdk/moonshotai": "^3.0.15",
         "@ai-sdk/openai": "^4.0.11",
         "@ai-sdk/openai-compatible": "^3.0.7",
         "@ai-sdk/xai": "^4.0.10",
@@ -234,6 +235,8 @@
 
     "@ai-sdk/mcp": ["@ai-sdk/mcp@2.0.10", "", { "dependencies": { "@ai-sdk/provider": "4.0.3", "@ai-sdk/provider-utils": "5.0.7", "pkce-challenge": "^5.0.1" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-LP47CEk3e6BwCqXamvv4nKbgOWQ5z5T/qS/J2fola5h98KTy7JT7I1yw5LKtB89MZBLviYmEh98Ep4ovHOz/vA=="],
 
+    "@ai-sdk/moonshotai": ["@ai-sdk/moonshotai@3.0.15", "", { "dependencies": { "@ai-sdk/openai-compatible": "3.0.12", "@ai-sdk/provider": "4.0.3", "@ai-sdk/provider-utils": "5.0.11" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-R4+UEk1f1+lare2dQ1xK1vOJLXFmKo2wZ/d1wXTCZGOsls2I4KXEgl6oq1rB2PRT6NkvA5DK52fMA9DaeQZoYQ=="],
+
     "@ai-sdk/openai": ["@ai-sdk/openai@4.0.11", "", { "dependencies": { "@ai-sdk/provider": "4.0.3", "@ai-sdk/provider-utils": "5.0.7" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-2PLnVPBsdJusgquPVYgPWd3ox4lbUFkp7DVUSmM0lQ4VISQoNxdgo6TvFAHgwQn9wJ/ckVYVRLU7+BvHc7T8ww=="],
 
     "@ai-sdk/openai-compatible": ["@ai-sdk/openai-compatible@3.0.7", "", { "dependencies": { "@ai-sdk/provider": "4.0.3", "@ai-sdk/provider-utils": "5.0.7" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-y8RVpm8BZKNHDxlbNVt/GYmoBBqLx10m4d7bOzOnRD8V2aP3Cr6/g/bfubR4EkbW1ySrnPr62svZIeND2fdp3Q=="],
@@ -3788,6 +3791,10 @@
 
     "@ai-sdk/amazon-bedrock/@ai-sdk/openai": ["@ai-sdk/openai@4.0.10", "", { "dependencies": { "@ai-sdk/provider": "4.0.3", "@ai-sdk/provider-utils": "5.0.7" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-K2t3JAshgw3KQ4kf2mEUl2gLCdZVbh1b1+G1lIc6+6JJjrZsp+ErOqqs2a29+hR6UkLI22vU//24bgA61eAJ6g=="],
 
+    "@ai-sdk/moonshotai/@ai-sdk/openai-compatible": ["@ai-sdk/openai-compatible@3.0.12", "", { "dependencies": { "@ai-sdk/provider": "4.0.3", "@ai-sdk/provider-utils": "5.0.11" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-tN9BUb4jUGjqtbRPUsziLxASfogLNICcq/Qrwr55vpnzKvprV6Ukl8ynED2lZaNrAHU5U4zlGnyU/iuYrjQK6g=="],
+
+    "@ai-sdk/moonshotai/@ai-sdk/provider-utils": ["@ai-sdk/provider-utils@5.0.11", "", { "dependencies": { "@ai-sdk/provider": "4.0.3", "@standard-schema/spec": "^1.1.0", "@workflow/serde": "4.1.0", "eventsource-parser": "^3.0.8" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-7/96wE+ZsKB35iS9ASyllrE4Ym/EolXEB7AkuJ5FI++fmS85BVTAs77890C+1Z2jwHfBKjBQSBmsliOsAh0iFQ=="],
+
     "@aws-crypto/sha256-browser/@smithy/util-utf8": ["@smithy/util-utf8@2.3.0", "", { "dependencies": { "@smithy/util-buffer-from": "^2.2.0", "tslib": "^2.6.2" } }, "sha512-R8Rdn8Hy72KKcebgLiv8jQcQkXoLMOGGv5uI1/k0l+snqkOzQ1R0ChUBCxWMlBsFMekWjq0wRudIweFs7sKT5A=="],
 
     "@aws-crypto/util/@smithy/util-utf8": ["@smithy/util-utf8@2.3.0", "", { "dependencies": { "@smithy/util-buffer-from": "^2.2.0", "tslib": "^2.6.2" } }, "sha512-R8Rdn8Hy72KKcebgLiv8jQcQkXoLMOGGv5uI1/k0l+snqkOzQ1R0ChUBCxWMlBsFMekWjq0wRudIweFs7sKT5A=="],
diff --git a/docs/config/models.mdx b/docs/config/models.mdx
index 06c2dd1ba5..30a7748217 100644
--- a/docs/config/models.mdx
+++ b/docs/config/models.mdx
@@ -34,7 +34,7 @@ Mux ships with curated models kept up to date with the frontier. Use any custom
 | Grok Code Fast 1       | xai:grok-code-fast-1          | `grok-code`                                                  |         |
 | DeepSeek V4 Pro        | deepseek:deepseek-v4-pro      | `deepseek`, `deepseek-pro`, `deepseek-v4`, `deepseek-v4-pro` |         |
 | DeepSeek V4 Flash      | deepseek:deepseek-v4-flash    | `deepseek-flash`, `deepseek-v4-flash`                        |         |
-| Kimi K3                | openrouter:moonshotai/kimi-k3 | `kimi`, `k3`, `kimi-k3`                                      |         |
+| Kimi K3                | moonshotai:kimi-k3            | `kimi`, `k3`, `kimi-k3`                                      |         |
 
 {/* END KNOWN_MODELS_TABLE */}
 
diff --git a/docs/config/providers.mdx b/docs/config/providers.mdx
index 435e543874..404baf546d 100644
--- a/docs/config/providers.mdx
+++ b/docs/config/providers.mdx
@@ -23,6 +23,7 @@ Most providers only need an API key. The UI handles validation and shows which p
 | **Google**         | Gemini Pro, Flash           | [aistudio.google.com](https://aistudio.google.com/)     |
 | **xAI**            | Grok                        | [console.x.ai](https://console.x.ai/)                   |
 | **DeepSeek**       | DeepSeek Chat, Reasoner     | [platform.deepseek.com](https://platform.deepseek.com/) |
+| **Moonshot AI**    | Kimi K3                     | [platform.moonshot.ai](https://platform.moonshot.ai/)   |
 | **OpenRouter**     | 300+ models                 | [openrouter.ai](https://openrouter.ai/)                 |
 | **Ollama**         | Local models                | [ollama.com](https://ollama.com/) (no key needed)       |
 | **Bedrock**        | Claude via AWS              | AWS Console                                             |
@@ -42,6 +43,7 @@ Providers also read from environment variables as fallback:
 | xAI            | `XAI_API_KEY`                                      |
 | OpenRouter     | `OPENROUTER_API_KEY`                               |
 | DeepSeek       | `DEEPSEEK_API_KEY`                                 |
+| Moonshot AI    | `MOONSHOT_API_KEY`                                 |
 | github-copilot | `GITHUB_COPILOT_TOKEN`                             |
 | Bedrock        | `AWS_REGION` (credentials via AWS SDK chain)       |
 
diff --git a/package.json b/package.json
index 16b7102d8d..f4be91c0e8 100644
--- a/package.json
+++ b/package.json
@@ -54,6 +54,7 @@
     "@ai-sdk/deepseek": "^3.0.7",
     "@ai-sdk/google": "^4.0.11",
     "@ai-sdk/mcp": "^2.0.10",
+    "@ai-sdk/moonshotai": "^3.0.15",
     "@ai-sdk/openai": "^4.0.11",
     "@ai-sdk/openai-compatible": "^3.0.7",
     "@ai-sdk/xai": "^4.0.10",
diff --git a/scripts/gen_docs.ts b/scripts/gen_docs.ts
index ed914eb56b..c33024cad5 100644
--- a/scripts/gen_docs.ts
+++ b/scripts/gen_docs.ts
@@ -369,6 +369,7 @@ const PROVIDER_DISPLAY_NAMES: Record<string, string> = {
   google: "Google",
   xai: "xAI",
   deepseek: "DeepSeek",
+  moonshotai: "Moonshot AI",
   openrouter: "OpenRouter",
   bedrock: "Bedrock",
 };
diff --git a/src/browser/assets/icons/moonshotai.svg b/src/browser/assets/icons/moonshotai.svg
new file mode 100644
index 0000000000..f57c5bd0d4
--- /dev/null
+++ b/src/browser/assets/icons/moonshotai.svg
@@ -0,0 +1,4 @@
+<!-- Moonshot AI mark (simplified monochrome crescent for CSS styling) -->
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24">
+  <path d="M21 13A9 9 0 1 1 11 3a7.5 7.5 0 0 0 10 10Z"/>
+</svg>
diff --git a/src/browser/components/ProjectSidebar/ProjectSidebar.test.tsx b/src/browser/components/ProjectSidebar/ProjectSidebar.test.tsx
index 3cd5d6c8f7..13a19a952b 100644
--- a/src/browser/components/ProjectSidebar/ProjectSidebar.test.tsx
+++ b/src/browser/components/ProjectSidebar/ProjectSidebar.test.tsx
@@ -60,6 +60,7 @@ function installProviderIconSvgMocks() {
     "@/browser/assets/icons/openrouter.svg?react",
     "@/browser/assets/icons/ollama.svg?react",
     "@/browser/assets/icons/deepseek.svg?react",
+    "@/browser/assets/icons/moonshotai.svg?react",
     "@/browser/assets/icons/aws.svg?react",
     "@/browser/assets/icons/github.svg?react",
   ] as const;
diff --git a/src/browser/components/ProviderIcon/ProviderIcon.tsx b/src/browser/components/ProviderIcon/ProviderIcon.tsx
index 221f0c5de6..12fb039d71 100644
--- a/src/browser/components/ProviderIcon/ProviderIcon.tsx
+++ b/src/browser/components/ProviderIcon/ProviderIcon.tsx
@@ -7,6 +7,7 @@ import XAIIcon from "@/browser/assets/icons/xai.svg?react";
 import OpenRouterIcon from "@/browser/assets/icons/openrouter.svg?react";
 import OllamaIcon from "@/browser/assets/icons/ollama.svg?react";
 import DeepSeekIcon from "@/browser/assets/icons/deepseek.svg?react";
+import MoonshotAIIcon from "@/browser/assets/icons/moonshotai.svg?react";
 import AWSIcon from "@/browser/assets/icons/aws.svg?react";
 import GitHubIcon from "@/browser/assets/icons/github.svg?react";
 import { GatewayIcon } from "@/browser/components/icons/GatewayIcon/GatewayIcon";
@@ -27,6 +28,7 @@ const PROVIDER_ICONS: Partial<Record<ProviderName, React.FC>> = {
   google: GoogleIcon,
   xai: XAIIcon,
   deepseek: DeepSeekIcon,
+  moonshotai: MoonshotAIIcon,
   openrouter: OpenRouterIcon,
   bedrock: AWSIcon,
   ollama: OllamaIcon,
diff --git a/src/browser/features/Settings/Sections/ProvidersSection.tsx b/src/browser/features/Settings/Sections/ProvidersSection.tsx
index 4bf8560d72..008cfd5f7a 100644
--- a/src/browser/features/Settings/Sections/ProvidersSection.tsx
+++ b/src/browser/features/Settings/Sections/ProvidersSection.tsx
@@ -243,6 +243,7 @@ const PROVIDER_KEY_URLS: Partial<Record<ProviderName, string>> = {
   google: "https://aistudio.google.com/app/apikey",
   xai: "https://console.x.ai/team/default/api-keys",
   deepseek: "https://platform.deepseek.com/api_keys",
+  moonshotai: "https://platform.moonshot.ai/console/api-keys",
   openrouter: "https://openrouter.ai/settings/keys",
   // bedrock: AWS credential chain, no simple key URL
   // ollama: local service, no key needed
diff --git a/src/cli/run.ts b/src/cli/run.ts
index a4af0608e2..fa49ccd66b 100644
--- a/src/cli/run.ts
+++ b/src/cli/run.ts
@@ -540,7 +540,7 @@ async function main(): Promise<number> {
       config.saveProvidersConfig(providersFromEnv);
     } else {
       throw new Error(
-        "No provider credentials found. Configure providers.jsonc or set ANTHROPIC_API_KEY / OPENAI_API_KEY / OPENROUTER_API_KEY / GOOGLE_GENERATIVE_AI_API_KEY."
+        "No provider credentials found. Configure providers.jsonc or set ANTHROPIC_API_KEY / OPENAI_API_KEY / OPENROUTER_API_KEY / GOOGLE_GENERATIVE_AI_API_KEY / MOONSHOT_API_KEY."
       );
     }
   }
diff --git a/src/common/config/schemas/providersConfig.ts b/src/common/config/schemas/providersConfig.ts
index beed8dfb43..30aaaf3114 100644
--- a/src/common/config/schemas/providersConfig.ts
+++ b/src/common/config/schemas/providersConfig.ts
@@ -68,6 +68,7 @@ export const MuxGatewayProviderConfigSchema = BaseProviderConfigSchema.extend({
 
 export const GoogleProviderConfigSchema = BaseProviderConfigSchema;
 export const DeepSeekProviderConfigSchema = BaseProviderConfigSchema;
+export const MoonshotAIProviderConfigSchema = BaseProviderConfigSchema;
 export const OllamaProviderConfigSchema = BaseProviderConfigSchema;
 export const GitHubCopilotProviderConfigSchema = BaseProviderConfigSchema;
 
@@ -81,6 +82,7 @@ export const ProvidersConfigSchema = z
     "mux-gateway": MuxGatewayProviderConfigSchema.optional(),
     google: GoogleProviderConfigSchema.optional(),
     deepseek: DeepSeekProviderConfigSchema.optional(),
+    moonshotai: MoonshotAIProviderConfigSchema.optional(),
     ollama: OllamaProviderConfigSchema.optional(),
     "github-copilot": GitHubCopilotProviderConfigSchema.optional(),
   })
@@ -95,6 +97,7 @@ export type XAIProviderConfig = z.infer<typeof XAIProviderConfigSchema>;
 export type MuxGatewayProviderConfig = z.infer<typeof MuxGatewayProviderConfigSchema>;
 export type GoogleProviderConfig = z.infer<typeof GoogleProviderConfigSchema>;
 export type DeepSeekProviderConfig = z.infer<typeof DeepSeekProviderConfigSchema>;
+export type MoonshotAIProviderConfig = z.infer<typeof MoonshotAIProviderConfigSchema>;
 export type OllamaProviderConfig = z.infer<typeof OllamaProviderConfigSchema>;
 export type GitHubCopilotProviderConfig = z.infer<typeof GitHubCopilotProviderConfigSchema>;
 
diff --git a/src/common/constants/knownModels.test.ts b/src/common/constants/knownModels.test.ts
index 092a0b3725..4f413d2d1b 100644
--- a/src/common/constants/knownModels.test.ts
+++ b/src/common/constants/knownModels.test.ts
@@ -14,9 +14,9 @@ describe("Known Models Integration", () => {
     for (const [key, model] of Object.entries(KNOWN_MODELS)) {
       const modelId = model.providerModelId;
 
-      // xAI and OpenRouter models are provider-prefixed in token metadata.
+      // xAI and Moonshot models are provider-prefixed in token metadata.
       const lookupKey =
-        model.provider === "xai" || model.provider === "openrouter"
+        model.provider === "xai" || model.provider === "moonshotai"
           ? `${model.provider}/${modelId}`
           : modelId;
       if (!(lookupKey in modelsJson) && !(lookupKey in modelsExtra) && !(modelId in modelsExtra)) {
@@ -46,9 +46,9 @@ describe("Known Models Integration", () => {
     expect(MODEL_ABBREVIATIONS["gpt-5.5"]).toBeUndefined();
   });
 
-  test("kimi aliases resolve to the OpenRouter-scoped Kimi K3 model", () => {
-    expect(MODEL_ABBREVIATIONS.kimi).toBe("openrouter:moonshotai/kimi-k3");
-    expect(MODEL_ABBREVIATIONS.k3).toBe("openrouter:moonshotai/kimi-k3");
+  test("kimi aliases resolve to the direct Moonshot Kimi K3 model", () => {
+    expect(MODEL_ABBREVIATIONS.kimi).toBe("moonshotai:kimi-k3");
+    expect(MODEL_ABBREVIATIONS.k3).toBe("moonshotai:kimi-k3");
   });
 
   test("known model ids and aliases stay unique across the curated registry", () => {
diff --git a/src/common/constants/knownModels.ts b/src/common/constants/knownModels.ts
index 1542a311cf..5d03d45aaf 100644
--- a/src/common/constants/knownModels.ts
+++ b/src/common/constants/knownModels.ts
@@ -4,7 +4,7 @@
 
 import { formatModelDisplayName } from "../utils/ai/modelDisplay";
 
-type ModelProvider = "anthropic" | "openai" | "google" | "xai" | "deepseek" | "openrouter";
+type ModelProvider = "anthropic" | "openai" | "google" | "xai" | "deepseek" | "moonshotai";
 
 interface KnownModelDefinition {
   /** Provider identifier used by SDK factories */
@@ -212,11 +212,11 @@ const MODEL_DEFINITIONS = {
     tokenizerOverride: "deepseek/deepseek-v3.1",
   },
   // Kimi K3 - Moonshot AI's flagship open-weight multimodal reasoning model (released
-  // July 16, 2026; 1M context, text+image input; $3/M in, $15/M out via OpenRouter).
+  // July 16, 2026; 1M context, text+image input; $3/M in, $15/M out).
   // Bare `kimi` alias tracks Moonshot's flagship per the shortest-alias convention.
   KIMI_K3: {
-    provider: "openrouter",
-    providerModelId: "moonshotai/kimi-k3",
+    provider: "moonshotai",
+    providerModelId: "kimi-k3",
     aliases: ["kimi", "k3", "kimi-k3"],
     // K3's tokenizer isn't published in ai-tokenizer yet; reuse Kimi K2 (the newest
     // Moonshot tokenizer available) for approximate counting.
diff --git a/src/common/constants/providers.ts b/src/common/constants/providers.ts
index 950a6ca204..37a1bbe080 100644
--- a/src/common/constants/providers.ts
+++ b/src/common/constants/providers.ts
@@ -20,6 +20,7 @@ export type ProviderName =
   | "google"
   | "xai"
   | "deepseek"
+  | "moonshotai"
   | "openrouter"
   | "github-copilot"
   | "bedrock"
@@ -128,13 +129,20 @@ export const PROVIDER_DEFINITIONS = {
     requiresApiKey: true,
     kind: "direct",
   },
+  moonshotai: {
+    displayName: "Moonshot AI",
+    import: () => import("@ai-sdk/moonshotai"),
+    factoryName: "createMoonshotAI",
+    requiresApiKey: true,
+    kind: "direct",
+  },
   openrouter: {
     displayName: "OpenRouter",
     import: () => import("@openrouter/ai-sdk-provider"),
     factoryName: "createOpenRouter",
     requiresApiKey: true,
     kind: "gateway",
-    routes: ["anthropic", "openai", "google", "xai", "deepseek"],
+    routes: ["anthropic", "openai", "google", "xai", "deepseek", "moonshotai"],
     passthrough: false,
     toGatewayModelId: toSlashSeparatedGatewayModelId,
     fromGatewayModelId: fromSlashSeparatedGatewayModelId,
diff --git a/src/common/types/thinking.ts b/src/common/types/thinking.ts
index 79b0772e49..53c08b845e 100644
--- a/src/common/types/thinking.ts
+++ b/src/common/types/thinking.ts
@@ -305,9 +305,9 @@ export function openaiSupportsProMode(modelString: string): boolean {
 }
 
 /**
- * Kimi K3 (Moonshot AI via OpenRouter) always reasons and supports only the max
- * reasoning effort; the thinking policy and the OpenRouter provider-options
- * branch both key off this predicate.
+ * Kimi K3 (Moonshot AI) always reasons and supports only the max reasoning
+ * effort; the thinking policy and the Moonshot/OpenRouter provider-options
+ * branches key off this predicate.
  */
 export function isKimiK3Model(modelString: string): boolean {
   return stripModelProviderPrefixes(modelString) === "kimi-k3";
diff --git a/src/common/utils/ai/modelCapabilities.test.ts b/src/common/utils/ai/modelCapabilities.test.ts
index d5492e9585..d95811ab72 100644
--- a/src/common/utils/ai/modelCapabilities.test.ts
+++ b/src/common/utils/ai/modelCapabilities.test.ts
@@ -33,11 +33,13 @@ describe("getModelCapabilities", () => {
     expect(caps).not.toBeNull();
   });
 
-  it("reports image support for Kimi K3 via the OpenRouter gateway id", () => {
-    const caps = getModelCapabilities("openrouter:moonshotai/kimi-k3");
-    expect(caps).not.toBeNull();
-    expect(caps?.supportsVision).toBe(true);
-    expect(caps?.supportsPdfInput).toBe(false);
+  it("reports image support for Kimi K3 via direct and gateway ids", () => {
+    for (const model of ["moonshotai:kimi-k3", "openrouter:moonshotai/kimi-k3"]) {
+      const caps = getModelCapabilities(model);
+      expect(caps).not.toBeNull();
+      expect(caps?.supportsVision).toBe(true);
+      expect(caps?.supportsPdfInput).toBe(false);
+    }
   });
 
   it("infers PDF support for OpenAI vision models when models-extra omits the flag", () => {
diff --git a/src/common/utils/ai/providerOptions.test.ts b/src/common/utils/ai/providerOptions.test.ts
index 9c4600667d..ec52ac363a 100644
--- a/src/common/utils/ai/providerOptions.test.ts
+++ b/src/common/utils/ai/providerOptions.test.ts
@@ -1564,11 +1564,34 @@ describe("buildProviderOptions - Google", () => {
   });
 });
 
+describe("buildProviderOptions - Moonshot", () => {
+  test("sends the explicit max reasoning effort for direct Kimi K3", () => {
+    expect(buildProviderOptions("moonshotai:kimi-k3", "max")).toEqual({
+      moonshotai: { reasoningEffort: "max" },
+    });
+  });
+
+  test("emits no provider options for other Moonshot models", () => {
+    expect(buildProviderOptions("moonshotai:kimi-k2.5", "medium")).toEqual({});
+  });
+});
+
 describe("buildProviderOptions - OpenRouter", () => {
-  test("sends the explicit max effort for Kimi K3", () => {
+  test("sends the explicit max effort for OpenRouter-routed Kimi K3", () => {
     // `enabled: true` alone falls back to OpenRouter's default (medium) effort,
     // which K3 does not support, so the effort must be sent explicitly.
-    expect(buildProviderOptions("openrouter:moonshotai/kimi-k3", "max")).toEqual({
+    const result = buildProviderOptions(
+      "moonshotai:kimi-k3",
+      "max",
+      undefined,
+      undefined,
+      undefined,
+      undefined,
+      undefined,
+      undefined,
+      "openrouter"
+    );
+    expect(result).toEqual({
       openrouter: {
         reasoning: {
           enabled: true,
diff --git a/src/common/utils/ai/providerOptions.ts b/src/common/utils/ai/providerOptions.ts
index fb09d41b05..fc4564e879 100644
--- a/src/common/utils/ai/providerOptions.ts
+++ b/src/common/utils/ai/providerOptions.ts
@@ -63,6 +63,15 @@ type OpenAICompatibleGatewayProviderOptions = Pick<
   "reasoningEffort"
 >;
 
+/**
+ * Moonshot AI provider options (@ai-sdk/moonshotai). Kimi K3 accepts only the
+ * literal "max" reasoning effort, matching the SDK's typed options schema.
+ */
+interface MoonshotAIProviderOptions {
+  [key: string]: JSONValue | undefined;
+  reasoningEffort?: "max";
+}
+
 /**
  * Provider-specific options structure for AI SDK
  */
@@ -71,6 +80,7 @@ type ProviderOptions =
   | { openai: OpenAIResponsesProviderOptions }
   | { google: GoogleGenerativeAIProviderOptions }
   | { openrouter: OpenRouterReasoningOptions }
+  | { moonshotai: MoonshotAIProviderOptions }
   | { xai: XaiProviderOptions }
   | { "github-copilot": OpenAICompatibleGatewayProviderOptions }
   | Record<string, never>; // Empty object for unsupported providers
@@ -475,6 +485,20 @@ export function buildProviderOptions(
     return options;
   }
 
+  // Build Moonshot-specific options
+  if (formatProvider === "moonshotai") {
+    // Kimi K3 always reasons and supports only the max reasoning effort. Send it
+    // explicitly rather than relying on the API default.
+    if (isKimiK3Model(capabilityModel)) {
+      const options = {
+        moonshotai: { reasoningEffort: "max" },
+      } satisfies { moonshotai: MoonshotAIProviderOptions };
+      log.debug("buildProviderOptions: Returning Moonshot options", options);
+      return options;
+    }
+    return {};
+  }
+
   // Build OpenRouter-specific options
   if (formatProvider === "openrouter") {
     // Kimi K3 always reasons and supports only the max reasoning effort. Send it
diff --git a/src/common/utils/thinking/policy.test.ts b/src/common/utils/thinking/policy.test.ts
index 7f76f2d763..42385bf276 100644
--- a/src/common/utils/thinking/policy.test.ts
+++ b/src/common/utils/thinking/policy.test.ts
@@ -608,10 +608,11 @@ describe("getThinkingPolicyForModel", () => {
     ]);
   });
 
-  test("returns a fixed max policy for Kimi K3 behind OpenRouter", () => {
+  test("returns a fixed max policy for Kimi K3 (direct and via OpenRouter)", () => {
+    expect(getThinkingPolicyForModel("moonshotai:kimi-k3")).toEqual(["max"]);
     expect(getThinkingPolicyForModel("openrouter:moonshotai/kimi-k3")).toEqual(["max"]);
     // Variant ids must not inherit the fixed K3 policy.
-    expect(getThinkingPolicyForModel("openrouter:moonshotai/kimi-k3-turbo")).toEqual([
+    expect(getThinkingPolicyForModel("moonshotai:kimi-k3-turbo")).toEqual([
       "off",
       "low",
       "medium",
diff --git a/src/common/utils/thinking/policy.ts b/src/common/utils/thinking/policy.ts
index 0fd2b2662a..e564896c46 100644
--- a/src/common/utils/thinking/policy.ts
+++ b/src/common/utils/thinking/policy.ts
@@ -169,8 +169,8 @@ function getExplicitThinkingPolicy(modelString: string): ThinkingPolicy | null {
     return ["low", "high"];
   }
 
-  // Kimi K3 always reasons and OpenRouter currently exposes only the max (default)
-  // reasoning effort, so the policy is a fixed single level.
+  // Kimi K3 always reasons and supports only the max reasoning effort, so the
+  // policy is a fixed single level.
   if (isKimiK3Model(withoutProviderNamespace)) {
     return ["max"];
   }
diff --git a/src/common/utils/tokens/modelStats.test.ts b/src/common/utils/tokens/modelStats.test.ts
index de4103c64e..17e99ccb8b 100644
--- a/src/common/utils/tokens/modelStats.test.ts
+++ b/src/common/utils/tokens/modelStats.test.ts
@@ -145,13 +145,16 @@ describe("getModelStats", () => {
     expect(flash.cache_read_input_token_cost).toBe(0.000000014);
   });
 
-  test("resolves Kimi K3 pricing and limits via the OpenRouter gateway id", () => {
-    const stats = expectStats("openrouter:moonshotai/kimi-k3");
+  test("resolves Kimi K3 pricing and limits via direct and gateway forms", () => {
+    const stats = expectStats("moonshotai:kimi-k3");
     expect(stats.max_input_tokens).toBe(1_048_576);
     expect(stats.max_output_tokens).toBe(131_072);
     expect(stats.input_cost_per_token).toBe(0.000003);
     expect(stats.output_cost_per_token).toBe(0.000015);
     expect(stats.cache_read_input_token_cost).toBe(0.0000003);
+
+    // The legacy OpenRouter form canonicalizes back to the direct Moonshot entry.
+    expect(expectStats("openrouter:moonshotai/kimi-k3")).toEqual(stats);
   });
 
   test("resolves the default image generation model pricing", () => {
diff --git a/src/common/utils/tokens/models-extra.ts b/src/common/utils/tokens/models-extra.ts
index a54b8c440f..01cb3cf43e 100644
--- a/src/common/utils/tokens/models-extra.ts
+++ b/src/common/utils/tokens/models-extra.ts
@@ -543,15 +543,15 @@ export const modelsExtra: Record<string, ModelData> = {
     supports_response_schema: true,
   },
 
-  // Kimi K3 via OpenRouter - released July 16, 2026. 1M context, 128K max output,
-  // text+image input (per the OpenRouter model page).
-  "openrouter/moonshotai/kimi-k3": {
+  // Kimi K3 - released July 16, 2026. 1M context, 128K max output, text+image input.
+  // Keyed on the direct provider; OpenRouter-routed requests canonicalize to this id.
+  "moonshotai/kimi-k3": {
     max_input_tokens: 1048576,
     max_output_tokens: 131072,
     input_cost_per_token: 0.000003, // $3 per million input tokens
     output_cost_per_token: 0.000015, // $15 per million output tokens
     cache_read_input_token_cost: 0.0000003, // $0.30 per million cached input tokens
-    litellm_provider: "openrouter",
+    litellm_provider: "moonshotai",
     mode: "chat",
     supports_function_calling: true,
     supports_vision: true,
diff --git a/src/node/services/agentSkills/builtInSkillContent.generated.ts b/src/node/services/agentSkills/builtInSkillContent.generated.ts
index cd25b0be45..75044179f0 100644
--- a/src/node/services/agentSkills/builtInSkillContent.generated.ts
+++ b/src/node/services/agentSkills/builtInSkillContent.generated.ts
@@ -3188,7 +3188,7 @@ export const BUILTIN_SKILL_FILES: Record<string, Record<string, string>> = {
       "| Grok Code Fast 1       | xai:grok-code-fast-1          | `grok-code`                                                  |         |",
       "| DeepSeek V4 Pro        | deepseek:deepseek-v4-pro      | `deepseek`, `deepseek-pro`, `deepseek-v4`, `deepseek-v4-pro` |         |",
       "| DeepSeek V4 Flash      | deepseek:deepseek-v4-flash    | `deepseek-flash`, `deepseek-v4-flash`                        |         |",
-      "| Kimi K3                | openrouter:moonshotai/kimi-k3 | `kimi`, `k3`, `kimi-k3`                                      |         |",
+      "| Kimi K3                | moonshotai:kimi-k3            | `kimi`, `k3`, `kimi-k3`                                      |         |",
       "",
       "{/* END KNOWN_MODELS_TABLE */}",
       "",
@@ -3617,6 +3617,7 @@ export const BUILTIN_SKILL_FILES: Record<string, Record<string, string>> = {
       "| **Google**         | Gemini Pro, Flash           | [aistudio.google.com](https://aistudio.google.com/)     |",
       "| **xAI**            | Grok                        | [console.x.ai](https://console.x.ai/)                   |",
       "| **DeepSeek**       | DeepSeek Chat, Reasoner     | [platform.deepseek.com](https://platform.deepseek.com/) |",
+      "| **Moonshot AI**    | Kimi K3                     | [platform.moonshot.ai](https://platform.moonshot.ai/)   |",
       "| **OpenRouter**     | 300+ models                 | [openrouter.ai](https://openrouter.ai/)                 |",
       "| **Ollama**         | Local models                | [ollama.com](https://ollama.com/) (no key needed)       |",
       "| **Bedrock**        | Claude via AWS              | AWS Console                                             |",
@@ -3636,6 +3637,7 @@ export const BUILTIN_SKILL_FILES: Record<string, Record<string, string>> = {
       "| xAI            | `XAI_API_KEY`                                      |",
       "| OpenRouter     | `OPENROUTER_API_KEY`                               |",
       "| DeepSeek       | `DEEPSEEK_API_KEY`                                 |",
+      "| Moonshot AI    | `MOONSHOT_API_KEY`                                 |",
       "| github-copilot | `GITHUB_COPILOT_TOKEN`                             |",
       "| Bedrock        | `AWS_REGION` (credentials via AWS SDK chain)       |",
       "",
diff --git a/src/node/utils/providerRequirements.ts b/src/node/utils/providerRequirements.ts
index 2d74204c40..40d5e19ef9 100644
--- a/src/node/utils/providerRequirements.ts
+++ b/src/node/utils/providerRequirements.ts
@@ -65,6 +65,9 @@ export const PROVIDER_ENV_VARS: Partial<
   deepseek: {
     apiKey: ["DEEPSEEK_API_KEY"],
   },
+  moonshotai: {
+    apiKey: ["MOONSHOT_API_KEY"],
+  },
   "github-copilot": {
     apiKey: ["GITHUB_COPILOT_TOKEN"],
   },

From eef38f90f4fe3b9f97eb229a67b85d51743b82c4 Mon Sep 17 00:00:00 2001
From: Michael Suchacz <203725896+ibetitsmike@users.noreply.github.com>
Date: Fri, 17 Jul 2026 18:45:01 +0000
Subject: [PATCH 4/4] ci: update flake offline cache hash for
 @ai-sdk/moonshotai

---
 flake.nix | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/flake.nix b/flake.nix
index 7ce2f87a80..3db1e58a37 100644
--- a/flake.nix
+++ b/flake.nix
@@ -83,7 +83,7 @@
 
             outputHashMode = "recursive";
             # Marker used by scripts/update_flake_hash.sh to update this hash in place.
-            outputHash = "sha256-PKc0nF1lp6yjPlveIhOc/Hp6DwFe3WgtfHyHTDQdUnQ="; # mux-offline-cache-hash
+            outputHash = "sha256-OzBNhFDq9TWcHNfEGI5SpOVvin9WVzi9SvsG34G267A="; # mux-offline-cache-hash
           };
 
           configurePhase = ''