diff --git a/CHANGELOG.md b/CHANGELOG.md
index 5909c9899..8a5cbd40f 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -6,6 +6,7 @@ Docs: https://docs.molt.bot
 Status: beta.
 
 ### Changes
+- Agents: add per-model fallbacks support in `agents.defaults.models[...].fallbacks` to configure model-specific fallback chains.
 - Rebrand: rename the npm package/CLI to `moltbot`, add a `moltbot` compatibility shim, and move extensions to the `@moltbot/*` scope.
 - Commands: group /help and /commands output with Telegram paging. (#2504) Thanks @hougangdev.
 - macOS: limit project-local `node_modules/.bin` PATH preference to debug builds (reduce PATH hijacking risk).
diff --git a/src/agents/model-fallback.test.ts b/src/agents/model-fallback.test.ts
index f8a94560c..a7e5f868c 100644
--- a/src/agents/model-fallback.test.ts
+++ b/src/agents/model-fallback.test.ts
@@ -538,4 +538,154 @@ describe("runWithModelFallback", () => {
     expect(result.provider).toBe("openai");
     expect(result.model).toBe("gpt-4.1-mini");
   });
+
+  it("uses per-model fallbacks from model catalog entry", async () => {
+    const cfg = {
+      agents: {
+        defaults: {
+          model: { primary: "openai/gpt-4.1-mini" },
+          models: {
+            "openai/gpt-4.1-mini": {
+              alias: "GPT-4",
+              fallbacks: ["anthropic/claude-haiku-3-5", "openai/gpt-4o"],
+            },
+            "anthropic/claude-haiku-3-5": {},
+            "openai/gpt-4o": {},
+          },
+        },
+      },
+    } as MoltbotConfig;
+
+    const run = vi
+      .fn()
+      .mockRejectedValueOnce(Object.assign(new Error("rate limit"), { status: 429 }))
+      .mockResolvedValueOnce("ok");
+
+    const result = await runWithModelFallback({
+      cfg,
+      provider: "openai",
+      model: "gpt-4.1-mini",
+      run,
+    });
+
+    expect(result.result).toBe("ok");
+    expect(result.provider).toBe("anthropic");
+    expect(result.model).toBe("claude-haiku-3-5");
+    expect(run).toHaveBeenCalledTimes(2);
+  });
+
+  it("per-model fallbacks take precedence over global fallbacks", async () => {
+    const cfg = {
+      agents: {
+        defaults: {
+          model: {
+            primary: "openrouter/z-ai/glm-4.5-air:free",
+            fallbacks: ["anthropic/claude-opus-4"],
+          },
+          models: {
+            "openrouter/z-ai/glm-4.5-air:free": {
+              alias: "GLM-Air",
+              fallbacks: ["zai/glm-4.5-air", "anthropic/claude-haiku-3-5"],
+            },
+            "zai/glm-4.5-air": {},
+            "anthropic/claude-haiku-3-5": {},
+            "anthropic/claude-opus-4": {},
+          },
+        },
+      },
+    } as MoltbotConfig;
+
+    const calls: Array<{ provider: string; model: string }> = [];
+
+    const result = await runWithModelFallback({
+      cfg,
+      provider: "openrouter",
+      model: "z-ai/glm-4.5-air:free",
+      run: async (provider, model) => {
+        calls.push({ provider, model });
+        if (calls.length === 1) {
+          throw Object.assign(new Error("rate limit"), { status: 429 });
+        }
+        if (calls.length === 2) {
+          throw Object.assign(new Error("auth error"), { status: 401 });
+        }
+        return "ok";
+      },
+    });
+
+    expect(result.result).toBe("ok");
+    expect(result.provider).toBe("anthropic");
+    expect(result.model).toBe("claude-haiku-3-5");
+    expect(calls).toEqual([
+      { provider: "openrouter", model: "z-ai/glm-4.5-air:free" },
+      { provider: "zai", model: "glm-4.5-air" },
+      { provider: "anthropic", model: "claude-haiku-3-5" },
+    ]);
+  });
+
+  it("empty per-model fallbacks array disables global fallbacks for that model", async () => {
+    const cfg = {
+      agents: {
+        defaults: {
+          model: {
+            primary: "openai/gpt-4.1-mini",
+            fallbacks: ["anthropic/claude-haiku-3-5"],
+          },
+          models: {
+            "openai/gpt-4.1-mini": {
+              fallbacks: [],
+            },
+          },
+        },
+      },
+    } as MoltbotConfig;
+
+    const calls: Array<{ provider: string; model: string }> = [];
+
+    await expect(
+      runWithModelFallback({
+        cfg,
+        provider: "openai",
+        model: "gpt-4.1-mini",
+        run: async (provider, model) => {
+          calls.push({ provider, model });
+          throw Object.assign(new Error("rate limit"), { status: 429 });
+        },
+      }),
+    ).rejects.toThrow("rate limit");
+
+    expect(calls).toEqual([{ provider: "openai", model: "gpt-4.1-mini" }]);
+  });
+
+  it("per-model fallbacks respect allowlist", async () => {
+    const cfg = {
+      agents: {
+        defaults: {
+          model: { primary: "openai/gpt-4.1-mini" },
+          models: {
+            "openai/gpt-4.1-mini": {
+              fallbacks: ["anthropic/claude-haiku-3-5"],
+            },
+            "anthropic/claude-haiku-3-5": {},
+          },
+        },
+      },
+    } as MoltbotConfig;
+
+    const run = vi
+      .fn()
+      .mockRejectedValueOnce(Object.assign(new Error("rate limit"), { status: 429 }))
+      .mockResolvedValueOnce("ok");
+
+    const result = await runWithModelFallback({
+      cfg,
+      provider: "openai",
+      model: "gpt-4.1-mini",
+      run,
+    });
+
+    expect(result.result).toBe("ok");
+    expect(result.provider).toBe("anthropic");
+    expect(result.model).toBe("claude-haiku-3-5");
+  });
 });
diff --git a/src/agents/model-fallback.ts b/src/agents/model-fallback.ts
index b87135f9f..f16be481e 100644
--- a/src/agents/model-fallback.ts
+++ b/src/agents/model-fallback.ts
@@ -165,11 +165,18 @@ function resolveFallbackCandidates(params: {
 
   const modelFallbacks = (() => {
     if (params.fallbacksOverride !== undefined) return params.fallbacksOverride;
-    const model = params.cfg?.agents?.defaults?.model as
+
+    // Check for per-model fallbacks first
+    const key = modelKey(provider, model);
+    const modelConfig = params.cfg?.agents?.defaults?.models?.[key];
+    if (modelConfig?.fallbacks) return modelConfig.fallbacks;
+
+    // Fall back to global model.fallbacks
+    const modelCfg = params.cfg?.agents?.defaults?.model as
       | { fallbacks?: string[] }
       | string
       | undefined;
-    if (model && typeof model === "object") return model.fallbacks ?? [];
+    if (modelCfg && typeof modelCfg === "object") return modelCfg.fallbacks ?? [];
     return [];
   })();
 
diff --git a/src/config/types.agent-defaults.ts b/src/config/types.agent-defaults.ts
index 9c6ce0211..189b7afde 100644
--- a/src/config/types.agent-defaults.ts
+++ b/src/config/types.agent-defaults.ts
@@ -16,6 +16,8 @@ export type AgentModelEntryConfig = {
   alias?: string;
   /** Provider-specific API parameters (e.g., GLM-4.7 thinking mode). */
   params?: Record<string, unknown>;
+  /** Per-model fallbacks (tried before global fallbacks). */
+  fallbacks?: string[];
 };
 
 export type AgentModelListConfig = {
diff --git a/src/config/zod-schema.agent-defaults.ts b/src/config/zod-schema.agent-defaults.ts
index a849078ed..396bfa6dd 100644
--- a/src/config/zod-schema.agent-defaults.ts
+++ b/src/config/zod-schema.agent-defaults.ts
@@ -37,6 +37,8 @@ export const AgentDefaultsSchema = z
             alias: z.string().optional(),
             /** Provider-specific API parameters (e.g., GLM-4.7 thinking mode). */
             params: z.record(z.string(), z.unknown()).optional(),
+            /** Per-model fallbacks (tried before global fallbacks). */
+            fallbacks: z.array(z.string()).optional(),
           })
           .strict(),
       )