diff --git a/packages/ai/test/anthropic-adaptive-thinking-models.test.ts b/packages/ai/test/anthropic-adaptive-thinking-models.test.ts
index 57badd3a1..1b1be72dc 100644
--- a/packages/ai/test/anthropic-adaptive-thinking-models.test.ts
+++ b/packages/ai/test/anthropic-adaptive-thinking-models.test.ts
@@ -8,7 +8,7 @@ const EXPECTED_CURRENT_ADAPTIVE_THINKING_MODELS = [
 	"anthropic/claude-opus-5",
 	"anthropic/claude-sonnet-5",
 	"cloudflare-ai-gateway/claude-fable-5",
-	"fireworks/accounts/fireworks/models/deepseek-v4-flash-0731",
+	"fireworks/accounts/fireworks/models/deepseek-v4p1-flash",
 	"fireworks/accounts/fireworks/models/gpt-oss-120b",
 	"fireworks/accounts/fireworks/models/qwen3p8-max",
 	"kimi-coding/kimi-for-coding",
diff --git a/packages/ai/test/anthropic-empty-thinking-signature-compat.test.ts b/packages/ai/test/anthropic-empty-thinking-signature-compat.test.ts
index 4811e895d..b15cdf4bc 100644
--- a/packages/ai/test/anthropic-empty-thinking-signature-compat.test.ts
+++ b/packages/ai/test/anthropic-empty-thinking-signature-compat.test.ts
@@ -106,12 +106,10 @@ describe("Anthropic empty thinking signature compat", () => {
 
 	// Regression for #9323: Fireworks emits unsigned thinking that must survive replay.
 	it.each([
-		"accounts/fireworks/models/deepseek-v4-flash-0731",
-		"accounts/fireworks/models/deepseek-v4-flash-vision-exp",
-		"accounts/fireworks/models/deepseek-v4-pro-0813",
+		"accounts/fireworks/models/deepseek-v4p1-flash",
 		"accounts/fireworks/models/qwen3p8-max",
 		"accounts/fireworks/models/qwen3p8-2p4t-a95b",
-		"accounts/fireworks/models/kimi-k2p6",
+		"accounts/fireworks/models/nemotron-3-ultra-nvfp4",
 	] as const)("preserves unsigned thinking for Fireworks %s", async (modelId) => {
 		const model = getModel("fireworks", modelId);
 		expect(model.compat?.allowEmptySignature).toBe(true);
@@ -127,10 +125,10 @@ describe("Anthropic empty thinking signature compat", () => {
 
 	// Regression for #9323: opting into unsigned replay must not change cross-model conversion.
 	it("still converts cross-model Fireworks thinking to text", async () => {
-		const model = getModel("fireworks", "accounts/fireworks/models/deepseek-v4-flash-0731");
+		const model = getModel("fireworks", "accounts/fireworks/models/deepseek-v4p1-flash");
 		const payload = await capturePayload(
 			model,
-			makeContext("", "internal reasoning", "fireworks", "accounts/fireworks/models/kimi-k2p6"),
+			makeContext("", "internal reasoning", "fireworks", "accounts/fireworks/models/nemotron-3-ultra-nvfp4"),
 		);
 		expect(payload.messages?.find((message) => message.role === "assistant")?.content).toEqual([
 			{ type: "text", text: "internal reasoning" },
diff --git a/packages/ai/test/fireworks-models.test.ts b/packages/ai/test/fireworks-models.test.ts
index f7661fa13..e3d4a3c8d 100644
--- a/packages/ai/test/fireworks-models.test.ts
+++ b/packages/ai/test/fireworks-models.test.ts
@@ -19,8 +19,8 @@ afterEach(() => {
 });
 
 describe("Fireworks models", () => {
-	it("registers the default Kimi K2.6 model via Anthropic-compatible Messages API", () => {
-		const model = getModel("fireworks", "accounts/fireworks/models/kimi-k2p6");
+	it("registers non-GLM, non-Kimi-K3 models via Anthropic-compatible Messages API", () => {
+		const model = getModel("fireworks", "accounts/fireworks/models/deepseek-v4p1-flash");
 
 		expect(model).toBeDefined();
 		expect(model.api).toBe("anthropic-messages");
@@ -28,19 +28,11 @@ describe("Fireworks models", () => {
 		expect(model.baseUrl).toBe("https://api.fireworks.ai/inference");
 		expect(model.reasoning).toBe(true);
 		expect(model.input).toEqual(["text", "image"]);
-		expect(model.contextWindow).toBe(262000);
-		expect(model.maxTokens).toBe(262000);
-		expect(model.cost).toEqual({
-			input: 0.95,
-			output: 4,
-			cacheRead: 0.16,
-			cacheWrite: 0,
-		});
 	});
 
-	it("aligns GLM 5.2 Fast with GLM 5.2's OpenAI-compatible config", () => {
-		const base = getModel("fireworks", "accounts/fireworks/models/glm-5p2");
-		const fast = getModel("fireworks", "accounts/fireworks/routers/glm-5p2-fast");
+	it("aligns GLM 5.3 Fast with GLM 5.3's OpenAI-compatible config", () => {
+		const base = getModel("fireworks", "accounts/fireworks/models/glm-5p3");
+		const fast = getModel("fireworks", "accounts/fireworks/routers/glm-5p3-fast");
 
 		expect(fast.api).toBe(base.api);
 		expect(fast.baseUrl).toBe(base.baseUrl);
@@ -48,7 +40,7 @@ describe("Fireworks models", () => {
 		expect(fast.thinkingLevelMap).toEqual(base.thinkingLevelMap);
 	});
 
-	it.each(["accounts/fireworks/models/glm-5p2", "accounts/fireworks/routers/glm-5p2-fast"] as const)(
+	it.each(["accounts/fireworks/models/glm-5p3", "accounts/fireworks/routers/glm-5p3-fast"] as const)(
 		"omits unsupported long cache retention for %s",
 		async (modelId) => {
 			const model = getModel("fireworks", modelId);
@@ -126,9 +118,7 @@ describe("Fireworks models", () => {
 
 	// Regression for #9323: native effort must reach Messages without budget-based fallback.
 	it.each([
-		["accounts/fireworks/models/deepseek-v4-flash-0731", ["off", "low", "high", "max"]],
-		["accounts/fireworks/models/deepseek-v4-flash-vision-exp", ["off", "low", "high", "max"]],
-		["accounts/fireworks/models/deepseek-v4-pro-0813", ["off", "low", "high", "max"]],
+		["accounts/fireworks/models/deepseek-v4p1-flash", ["off", "low", "high", "max"]],
 		["accounts/fireworks/models/qwen3p8-max", ["off", "low", "medium", "xhigh"]],
 		["accounts/fireworks/models/qwen3p8-2p4t-a95b", ["off", "low", "medium", "xhigh"]],
 	] as const)("sends native Messages effort levels for %s", async (modelId, levels) => {
@@ -161,8 +151,8 @@ describe("Fireworks models", () => {
 
 	// Regression for #9323: accepted aliases are not distinct native effort levels.
 	it.each([
-		["accounts/fireworks/models/glm-5p2", ["off", "high", "max"]],
-		["accounts/fireworks/routers/glm-5p2-fast", ["off", "high", "max"]],
+		["accounts/fireworks/models/glm-5p3", ["low", "high", "max"]],
+		["accounts/fireworks/routers/glm-5p3-fast", ["low", "high", "max"]],
 		["accounts/fireworks/models/kimi-k3", ["low", "high", "max"]],
 		["accounts/fireworks/routers/kimi-k3-fast", ["low", "high", "max"]],
 	] as const)("exposes distinct native effort levels for %s", (modelId, levels) => {
@@ -170,7 +160,7 @@ describe("Fireworks models", () => {
 	});
 
 	it("keeps toggle-only Messages models without a verified fallback on budget-based thinking", async () => {
-		const model = getModel("fireworks", "accounts/fireworks/models/kimi-k2p6");
+		const model = getModel("fireworks", "accounts/fireworks/models/nemotron-3-ultra-nvfp4");
 		expect(model.compat?.forceAdaptiveThinking).toBeUndefined();
 		let payload: Record<string, unknown> | undefined;
 		await streamSimple(
@@ -197,7 +187,7 @@ describe("Fireworks models", () => {
 	});
 
 	it("sets Fireworks-specific compat for session affinity and unsupported tool fields", () => {
-		const model = getModel("fireworks", "accounts/fireworks/models/kimi-k2p6");
+		const model = getModel("fireworks", "accounts/fireworks/models/nemotron-3-ultra-nvfp4");
 
 		expect(model.compat).toBeDefined();
 		expect(model.compat?.sendSessionAffinityHeaders).toBe(true);
diff --git a/packages/ai/test/openai-completions-prompt-cache.test.ts b/packages/ai/test/openai-completions-prompt-cache.test.ts
index 0beda273e..45f3df5fa 100644
--- a/packages/ai/test/openai-completions-prompt-cache.test.ts
+++ b/packages/ai/test/openai-completions-prompt-cache.test.ts
@@ -171,7 +171,7 @@ describe("openai-completions prompt caching", () => {
 		expect(headers["x-session-affinity"]).toBe("session-affinity");
 	});
 
-	it.each(["accounts/fireworks/models/glm-5p2", "accounts/fireworks/routers/glm-5p2-fast"] as const)(
+	it.each(["accounts/fireworks/models/glm-5p3", "accounts/fireworks/routers/glm-5p3-fast"] as const)(
 		"sends Fireworks session affinity for %s",
 		async (modelId) => {
 			const model = getModel("fireworks", modelId);
diff --git a/packages/ai/test/providers.test.ts b/packages/ai/test/providers.test.ts
index 54a8c7490..41309b28c 100644
--- a/packages/ai/test/providers.test.ts
+++ b/packages/ai/test/providers.test.ts
@@ -192,7 +192,7 @@ describe("builtin providers", () => {
 			["openrouter", "openai/gpt-5.6-terra"],
 		] as const;
 		const unsupported = [
-			["fireworks", "accounts/fireworks/models/kimi-k2p6"],
+			["fireworks", "accounts/fireworks/models/nemotron-3-ultra-nvfp4"],
 			["openai", "gpt-4.1"],
 			["openai", "gpt-5.2"],
 			["anthropic", "claude-sonnet-4-5"],
diff --git a/packages/coding-agent/CHANGELOG.md b/packages/coding-agent/CHANGELOG.md
index d90632a72..488bb8fa1 100644
--- a/packages/coding-agent/CHANGELOG.md
+++ b/packages/coding-agent/CHANGELOG.md
@@ -24,6 +24,7 @@
 - Fixed `RpcClient` skipping the next event listener when a listener unsubscribes while handling an event, which could make `waitForIdle()` time out after `collectEvents()` ([#9990](https://github.com/earendil-works/pi/issues/9990)).
 - Fixed full-file `read` calls rendering as `:1` when models send `null` for omitted `offset` and `limit` ([#9996](https://github.com/earendil-works/pi/issues/9996)).
 - Fixed new sessions being lost when pi exits before the first assistant response. The session file is now created when the first user message is sent ([#10000](https://github.com/earendil-works/pi/issues/10000)).
+- Fixed the Fireworks default model pointing at the removed Kimi K2.6 model; it now defaults to Kimi K3.
 
 ## [0.87.1] - 2026-09-22
 
diff --git a/packages/coding-agent/src/core/model-resolver.ts b/packages/coding-agent/src/core/model-resolver.ts
index d4a4805fb..678527310 100644
--- a/packages/coding-agent/src/core/model-resolver.ts
+++ b/packages/coding-agent/src/core/model-resolver.ts
@@ -43,7 +43,7 @@ export const defaultModelPerProvider: Partial<Record<KnownProvider, string>> = {
 	moonshotai: "kimi-k2.6",
 	"moonshotai-cn": "kimi-k2.6",
 	huggingface: "moonshotai/Kimi-K2.6",
-	fireworks: "accounts/fireworks/models/kimi-k2p6",
+	fireworks: "accounts/fireworks/models/kimi-k3",
 	together: "moonshotai/Kimi-K2.6",
 	baseten: "zai-org/GLM-5.2",
 	opencode: "kimi-k2.6",