diff --git a/CHANGELOG.md b/CHANGELOG.md
index 51be42c34df..af678994d88 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -82,6 +82,7 @@ Docs: https://docs.openclaw.ai
 - Agents/sessions: persist the runtime-resolved context budget from embedded agent runs, so Codex GPT-5.5 sessions keep the catalog/runtime context cap instead of falling back to the generic 200k status value. Fixes #71294. Thanks @tud0r.
 - Agents/tools: fail runs before model submission when explicit tool allowlists resolve to no callable tools, preventing text-only hallucinated tool results for missing tools such as plugin commands that were not registered. Fixes #71292. Thanks @steipete.
 - Agents/embedded: skip provider submission when an embedded run has no prompt, replay history, or prompt-local images, preventing empty OpenAI Responses requests from surfacing provider errors into user channels. Fixes #71130. Thanks @steipete.
+- Providers/Google: map `/think adaptive` to Gemini dynamic thinking instead of a fixed medium/high budget, using Gemini 3's provider default and Gemini 2.5's `thinkingBudget: -1`. Fixes #71316. Thanks @steipete.
 - Providers/MiniMax: keep M2.7 chat model metadata text-only so image tool requests route through `MiniMax-VL-01` instead of the Anthropic-compatible chat endpoint. Fixes #71296. Thanks @ilker-cevikkaya.
 - Discord/replies: run `message_sending` plugin hooks for Discord reply delivery, including DM targets, so plugins can transform or cancel outbound Discord replies consistently with other channels. Fixes #59350. (#71094) Thanks @wei840222.
 - Control UI/commands: carry provider-owned thinking option ids/labels in session rows and defaults so fresh sessions show and accept dynamic modes such as `adaptive`, `xhigh`, and `max`. Fixes #71269. Thanks @Young-Khalil.
diff --git a/docs/tools/thinking.md b/docs/tools/thinking.md
index f7c8f694b6c..30faa6321ad 100644
--- a/docs/tools/thinking.md
+++ b/docs/tools/thinking.md
@@ -14,7 +14,7 @@ title: "Thinking levels"
   - medium → “think harder”
   - high → “ultrathink” (max budget)
   - xhigh → “ultrathink+” (GPT-5.2+ and Codex models, plus Anthropic Claude Opus 4.7 effort)
-  - adaptive → provider-managed adaptive thinking (supported for Claude 4.6 on Anthropic/Bedrock and Anthropic Claude Opus 4.7)
+  - adaptive → provider-managed adaptive thinking (supported for Claude 4.6 on Anthropic/Bedrock, Anthropic Claude Opus 4.7, and Google Gemini dynamic thinking)
   - max → provider max reasoning (currently Anthropic Claude Opus 4.7)
   - `x-high`, `x_high`, `extra-high`, `extra high`, and `extra_high` map to `xhigh`.
   - `highest` maps to `high`.
@@ -27,6 +27,7 @@ title: "Thinking levels"
   - Anthropic Claude Opus 4.7 maps `/think xhigh` to adaptive thinking plus `output_config.effort: "xhigh"`, because `/think` is a thinking directive and `xhigh` is the Opus 4.7 effort setting.
   - Anthropic Claude Opus 4.7 also exposes `/think max`; it maps to the same provider-owned max effort path.
   - OpenAI GPT models map `/think` through model-specific Responses API effort support. `/think off` sends `reasoning.effort: "none"` only when the target model supports it; otherwise OpenClaw omits the disabled reasoning payload instead of sending an unsupported value.
+  - Google Gemini maps `/think adaptive` to Gemini's provider-owned dynamic thinking. Gemini 3 requests omit a fixed `thinkingLevel`, while Gemini 2.5 requests send `thinkingBudget: -1`; fixed levels still map to the closest Gemini `thinkingLevel` or budget for that model family.
   - MiniMax (`minimax/*`) on the Anthropic-compatible streaming path defaults to `thinking: { type: "disabled" }` unless you explicitly set thinking in model params or request params. This avoids leaked `reasoning_content` deltas from MiniMax's non-native Anthropic stream format.
   - Z.AI (`zai/*`) only supports binary thinking (`on`/`off`). Any non-`off` level is treated as `on` (mapped to `low`).
   - Moonshot (`moonshot/*`) maps `/think off` to `thinking: { type: "disabled" }` and any non-`off` level to `thinking: { type: "enabled" }`. When thinking is enabled, Moonshot only accepts `tool_choice` `auto|none`; OpenClaw normalizes incompatible values to `auto`.
diff --git a/extensions/google/index.test.ts b/extensions/google/index.test.ts
index 66d14fe1fd0..400ef9cdf89 100644
--- a/extensions/google/index.test.ts
+++ b/extensions/google/index.test.ts
@@ -180,6 +180,40 @@ describe("google provider plugin hooks", () => {
     runCase(cliProvider, "google-gemini-cli");
   });
 
+  it("advertises adaptive thinking for Gemini dynamic thinking", async () => {
+    const { providers } = await registerProviderPlugin({
+      plugin: googleProviderPlugin,
+      id: "google",
+      name: "Google Provider",
+    });
+    const provider = requireRegisteredProvider(providers, "google");
+    expect(provider.resolveThinkingProfile).toBeDefined();
+    const resolveThinkingProfile = provider.resolveThinkingProfile!;
+    const gemini3Profile = resolveThinkingProfile({
+      provider: "google",
+      modelId: "gemini-3.1-pro-preview",
+    } as never);
+    const gemini25Profile = resolveThinkingProfile({
+      provider: "google",
+      modelId: "gemini-2.5-flash",
+    } as never);
+
+    expect(gemini3Profile?.levels).toEqual([
+      { id: "off" },
+      { id: "low" },
+      { id: "adaptive" },
+      { id: "high" },
+    ]);
+    expect(gemini25Profile?.levels).toEqual([
+      { id: "off" },
+      { id: "minimal" },
+      { id: "low" },
+      { id: "medium" },
+      { id: "adaptive" },
+      { id: "high" },
+    ]);
+  });
+
   it("shares Gemini replay and stream hooks across Google provider variants", async () => {
     const { providers } = await registerProviderPlugin({
       plugin: googleProviderPlugin,
diff --git a/extensions/google/provider-hooks.ts b/extensions/google/provider-hooks.ts
index a1a095b54d9..7100bcf3ea4 100644
--- a/extensions/google/provider-hooks.ts
+++ b/extensions/google/provider-hooks.ts
@@ -12,8 +12,15 @@ export const GOOGLE_GEMINI_PROVIDER_HOOKS = {
   resolveThinkingProfile: ({ modelId }: ProviderDefaultThinkingPolicyContext) =>
     ({
       levels: isGoogleGemini3ProModel(modelId)
-        ? [{ id: "off" }, { id: "low" }, { id: "high" }]
-        : [{ id: "off" }, { id: "minimal" }, { id: "low" }, { id: "medium" }, { id: "high" }],
+        ? [{ id: "off" }, { id: "low" }, { id: "adaptive" }, { id: "high" }]
+        : [
+            { id: "off" },
+            { id: "minimal" },
+            { id: "low" },
+            { id: "medium" },
+            { id: "adaptive" },
+            { id: "high" },
+          ],
     }) satisfies ProviderThinkingProfile,
   wrapStreamFn: createGoogleThinkingStreamWrapper,
 };
diff --git a/extensions/google/thinking-api.ts b/extensions/google/thinking-api.ts
index a13526e90cb..fbb47dfdc59 100644
--- a/extensions/google/thinking-api.ts
+++ b/extensions/google/thinking-api.ts
@@ -1,6 +1,7 @@
 export {
   createGoogleThinkingPayloadWrapper,
   createGoogleThinkingStreamWrapper,
+  isGoogleGemini25ThinkingBudgetModel,
   isGoogleGemini3FlashModel,
   isGoogleGemini3ProModel,
   isGoogleGemini3ThinkingLevelModel,
diff --git a/extensions/google/thinking.test.ts b/extensions/google/thinking.test.ts
index d05f5cbc9e5..bd4dadf49b7 100644
--- a/extensions/google/thinking.test.ts
+++ b/extensions/google/thinking.test.ts
@@ -10,7 +10,7 @@ describe("google thinking policy", () => {
     ["minimal", "LOW"],
     ["low", "LOW"],
     ["medium", "HIGH"],
-    ["adaptive", "HIGH"],
+    ["adaptive", undefined],
     ["high", "HIGH"],
     ["xhigh", "HIGH"],
   ] as const)("maps Gemini 3 Pro thinking level %s to %s", (thinkingLevel, expected) => {
@@ -40,7 +40,7 @@ describe("google thinking policy", () => {
     ["minimal", "MINIMAL"],
     ["low", "LOW"],
     ["medium", "MEDIUM"],
-    ["adaptive", "MEDIUM"],
+    ["adaptive", undefined],
     ["high", "HIGH"],
     ["xhigh", "HIGH"],
   ] as const)("maps Gemini 3 Flash thinking level %s to %s", (thinkingLevel, expected) => {
@@ -53,7 +53,7 @@ describe("google thinking policy", () => {
   });
 
   it.each([
-    [-1, "MINIMAL"],
+    [-1, undefined],
     [0, "MINIMAL"],
     [2048, "LOW"],
     [8192, "MEDIUM"],
@@ -98,6 +98,43 @@ describe("google thinking policy", () => {
     });
   });
 
+  it("keeps Gemini 3 adaptive thinking provider-dynamic instead of forcing a fixed level", () => {
+    const payload = {
+      generationConfig: {
+        thinkingConfig: { thinkingBudget: 8192, includeThoughts: true },
+      },
+    };
+
+    sanitizeGoogleThinkingPayload({
+      payload,
+      modelId: "gemini-3-flash-preview",
+      thinkingLevel: "adaptive",
+    });
+
+    expect(payload.generationConfig.thinkingConfig).toEqual({
+      includeThoughts: true,
+    });
+  });
+
+  it("maps Gemini 2.5 adaptive thinking to dynamic thinkingBudget", () => {
+    const payload = {
+      config: {
+        thinkingConfig: { thinkingBudget: 8192, includeThoughts: true },
+      },
+    };
+
+    sanitizeGoogleThinkingPayload({
+      payload,
+      modelId: "gemini-2.5-flash",
+      thinkingLevel: "adaptive",
+    });
+
+    expect(payload.config.thinkingConfig).toEqual({
+      includeThoughts: true,
+      thinkingBudget: -1,
+    });
+  });
+
   it("maps Gemma 4 thinking mode without sending thinkingBudget", () => {
     const payload = {
       config: {
diff --git a/extensions/google/thinking.ts b/extensions/google/thinking.ts
index 89aac457cd1..edc0735a5f6 100644
--- a/extensions/google/thinking.ts
+++ b/extensions/google/thinking.ts
@@ -1,6 +1,7 @@
 export {
   createGoogleThinkingPayloadWrapper,
   createGoogleThinkingStreamWrapper,
+  isGoogleGemini25ThinkingBudgetModel,
   isGoogleGemini3FlashModel,
   isGoogleGemini3ProModel,
   isGoogleGemini3ThinkingLevelModel,
diff --git a/extensions/google/transport-stream.test.ts b/extensions/google/transport-stream.test.ts
index 902febfef19..e118951c0e4 100644
--- a/extensions/google/transport-stream.test.ts
+++ b/extensions/google/transport-stream.test.ts
@@ -444,6 +444,44 @@ describe("google transport stream", () => {
     });
   });
 
+  it("keeps adaptive Gemini 3 thinking on provider dynamic defaults", () => {
+    const params = buildGoogleGenerativeAiParams(
+      buildGeminiModel({ id: "gemini-3-flash-preview" }),
+      {
+        messages: [{ role: "user", content: "hello", timestamp: 0 }],
+      } as never,
+      {
+        reasoning: "adaptive",
+      } as never,
+    );
+
+    expect(params.generationConfig).toMatchObject({
+      thinkingConfig: { includeThoughts: true },
+    });
+    expect(params.generationConfig).not.toMatchObject({
+      thinkingConfig: { thinkingLevel: expect.any(String) },
+    });
+    expect(params.generationConfig).not.toMatchObject({
+      thinkingConfig: { thinkingBudget: expect.any(Number) },
+    });
+  });
+
+  it("maps adaptive Gemini 2.5 thinking to dynamic thinkingBudget", () => {
+    const params = buildGoogleGenerativeAiParams(
+      buildGeminiModel({ id: "gemini-2.5-flash" }),
+      {
+        messages: [{ role: "user", content: "hello", timestamp: 0 }],
+      } as never,
+      {
+        reasoning: "adaptive",
+      } as never,
+    );
+
+    expect(params.generationConfig).toMatchObject({
+      thinkingConfig: { includeThoughts: true, thinkingBudget: -1 },
+    });
+  });
+
   it("normalizes explicit Gemini 3 Pro thinking levels", () => {
     const params = buildGoogleGenerativeAiParams(
       buildGeminiModel({ id: "gemini-3.1-pro-preview" }),
diff --git a/extensions/google/transport-stream.ts b/extensions/google/transport-stream.ts
index 1f1b6041f3c..6db1cb6f167 100644
--- a/extensions/google/transport-stream.ts
+++ b/extensions/google/transport-stream.ts
@@ -25,6 +25,7 @@ import { normalizeLowercaseStringOrEmpty } from "openclaw/plugin-sdk/text-runtim
 import { parseGeminiAuth } from "./gemini-auth.js";
 import { normalizeGoogleApiBaseUrl } from "./provider-policy.js";
 import {
+  isGoogleGemini25ThinkingBudgetModel,
   isGoogleGemini3FlashModel,
   isGoogleGemini3ProModel,
   resolveGoogleGemini3ThinkingLevel,
@@ -238,6 +239,10 @@ function getGoogleThinkingBudget(
   return undefined;
 }
 
+function isAdaptiveReasoningLevel(value: unknown): value is "adaptive" {
+  return value === "adaptive";
+}
+
 function resolveGoogleThinkingConfig(
   model: GoogleTransportModel,
   options: GoogleTransportOptions | undefined,
@@ -268,6 +273,17 @@ function resolveGoogleThinkingConfig(
   if (!options?.reasoning) {
     return getDisabledThinkingConfig(model.id);
   }
+  if (isAdaptiveReasoningLevel(options.reasoning)) {
+    if (isGoogleGemini3ProModel(model.id) || isGoogleGemini3FlashModel(model.id)) {
+      return { includeThoughts: true };
+    }
+    if (isGoogleGemini25ThinkingBudgetModel(model.id)) {
+      return normalizeGoogleThinkingConfig(model.id, {
+        includeThoughts: true,
+        thinkingBudget: -1,
+      });
+    }
+  }
   if (isGoogleGemini3ProModel(model.id) || isGoogleGemini3FlashModel(model.id)) {
     return {
       includeThoughts: true,
diff --git a/src/agents/pi-embedded-runner/google-stream-wrappers.test.ts b/src/agents/pi-embedded-runner/google-stream-wrappers.test.ts
index e7b4c8d58bf..44556ff6b8f 100644
--- a/src/agents/pi-embedded-runner/google-stream-wrappers.test.ts
+++ b/src/agents/pi-embedded-runner/google-stream-wrappers.test.ts
@@ -97,7 +97,7 @@ describe("sanitizeGoogleThinkingPayload — gemini-2.5-pro zero budget", () => {
     });
   });
 
-  it("fills thinkingLevel for Gemini 3 Flash negative budgets", () => {
+  it("rewrites Gemini 3 Flash negative budgets when a fixed thinking level is explicit", () => {
     const payload = {
       config: {
         thinkingConfig: { thinkingBudget: -1, includeThoughts: true },
@@ -113,4 +113,37 @@ describe("sanitizeGoogleThinkingPayload — gemini-2.5-pro zero budget", () => {
       thinkingLevel: "MEDIUM",
     });
   });
+
+  it("keeps Gemini 3 adaptive thinking on provider dynamic defaults", () => {
+    const payload = {
+      config: {
+        thinkingConfig: { thinkingBudget: 8192, includeThoughts: true },
+      },
+    };
+    sanitizeGoogleThinkingPayload({
+      payload,
+      modelId: "gemini-3-flash-preview",
+      thinkingLevel: "adaptive",
+    });
+    expect(payload.config.thinkingConfig).toEqual({
+      includeThoughts: true,
+    });
+  });
+
+  it("maps Gemini 2.5 adaptive thinking to thinkingBudget=-1", () => {
+    const payload = {
+      config: {
+        thinkingConfig: { thinkingBudget: 8192, includeThoughts: true },
+      },
+    };
+    sanitizeGoogleThinkingPayload({
+      payload,
+      modelId: "gemini-2.5-flash",
+      thinkingLevel: "adaptive",
+    });
+    expect(payload.config.thinkingConfig).toEqual({
+      includeThoughts: true,
+      thinkingBudget: -1,
+    });
+  });
 });
diff --git a/src/plugin-sdk/provider-stream-shared.ts b/src/plugin-sdk/provider-stream-shared.ts
index 9ebf3733574..34135faceb2 100644
--- a/src/plugin-sdk/provider-stream-shared.ts
+++ b/src/plugin-sdk/provider-stream-shared.ts
@@ -158,6 +158,10 @@ export function isGoogleThinkingRequiredModel(modelId: string): boolean {
   return normalizeLowercaseStringOrEmpty(modelId).includes("gemini-2.5-pro");
 }
 
+export function isGoogleGemini25ThinkingBudgetModel(modelId: string): boolean {
+  return /(?:^|\/)gemini-2\.5-/.test(normalizeLowercaseStringOrEmpty(modelId));
+}
+
 export function isGoogleGemini3ProModel(modelId: string): boolean {
   const normalized = normalizeLowercaseStringOrEmpty(modelId);
   return /(?:^|\/)gemini-(?:3(?:\.\d+)?-pro|pro-latest)(?:-|$)/.test(normalized);
@@ -187,15 +191,19 @@ export function resolveGoogleGemini3ThinkingLevel(params: {
       case "low":
         return "LOW";
       case "medium":
-      case "adaptive":
       case "high":
       case "max":
       case "xhigh":
         return "HIGH";
+      case "adaptive":
+        return undefined;
       case undefined:
         break;
     }
     if (typeof params.thinkingBudget === "number") {
+      if (params.thinkingBudget < 0) {
+        return undefined;
+      }
       return params.thinkingBudget <= 2048 ? "LOW" : "HIGH";
     }
     return undefined;
@@ -210,18 +218,22 @@ export function resolveGoogleGemini3ThinkingLevel(params: {
     case "low":
       return "LOW";
     case "medium":
-    case "adaptive":
       return "MEDIUM";
     case "high":
     case "max":
     case "xhigh":
       return "HIGH";
+    case "adaptive":
+      return undefined;
     case undefined:
       break;
   }
   if (typeof params.thinkingBudget !== "number") {
     return undefined;
   }
+  if (params.thinkingBudget < 0) {
+    return undefined;
+  }
   if (params.thinkingBudget <= 0) {
     return "MINIMAL";
   }
@@ -355,6 +367,29 @@ function sanitizeGoogleThinkingConfigContainer(params: {
 
   const thinkingBudget = thinkingConfigObj.thinkingBudget;
 
+  if (
+    params.thinkingLevel === "adaptive" &&
+    typeof params.modelId === "string" &&
+    isGoogleGemini25ThinkingBudgetModel(params.modelId)
+  ) {
+    delete thinkingConfigObj.thinkingLevel;
+    thinkingConfigObj.thinkingBudget = -1;
+    return;
+  }
+
+  if (
+    params.thinkingLevel === "adaptive" &&
+    typeof params.modelId === "string" &&
+    isGoogleGemini3ThinkingLevelModel(params.modelId)
+  ) {
+    delete thinkingConfigObj.thinkingBudget;
+    delete thinkingConfigObj.thinkingLevel;
+    if (Object.keys(thinkingConfigObj).length === 0) {
+      delete configObj.thinkingConfig;
+    }
+    return;
+  }
+
   if (typeof params.modelId === "string" && isGoogleGemini3ThinkingLevelModel(params.modelId)) {
     const mappedLevel = resolveGoogleGemini3ThinkingLevel({
       modelId: params.modelId,