diff --git a/packages/core/src/v1/config/config.ts b/packages/core/src/v1/config/config.ts
index 7ebb4b69b023..e8379788772c 100644
--- a/packages/core/src/v1/config/config.ts
+++ b/packages/core/src/v1/config/config.ts
@@ -182,6 +182,10 @@ export const Info = Schema.Struct({
       mcp_timeout: Schema.optional(PositiveInt).annotate({
         description: "Timeout in milliseconds for model context protocol (MCP) requests",
       }),
+      sampling_defaults: Schema.optional(Schema.Boolean).annotate({
+        description:
+          "Apply built-in sampling defaults (temperature, top_p, top_k) for known models. Enabled by default; set to false to leave sampling to provider defaults. Explicit per-agent values and chat.params plugins always apply.",
+      }),
       policies: Schema.optional(Schema.mutable(Schema.Array(ConfigExperimental.Policy))).annotate({
         description: "Policy statements applied to supported resources, such as provider access",
       }),
diff --git a/packages/opencode/src/provider/transform.ts b/packages/opencode/src/provider/transform.ts
index fdd03d520566..e9752be57e41 100644
--- a/packages/opencode/src/provider/transform.ts
+++ b/packages/opencode/src/provider/transform.ts
@@ -528,6 +528,9 @@ const GEMINI_MODELS_WITH_SAMPLING_DEFAULTS = [
 export function temperature(model: Provider.Model) {
   const id = model.api.id.toLowerCase()
   if (id.includes("north-mini-code")) return 1.0
+  if (id.includes("qwen3-coder-next") || id.includes("qwen3.8") || id.includes("qwen-3.8")) return 1.0
+  if (id.includes("qwen3") || id.includes("qwen-3")) return 0.6
+  if (id.includes("qwen2.5-coder")) return 0.7
   if (id.includes("qwen")) return 0.55
   if (id.includes("claude")) return undefined
   if (id.includes("gemini"))
@@ -547,6 +550,8 @@ export function temperature(model: Provider.Model) {
 
 export function topP(model: Provider.Model) {
   const id = model.api.id.toLowerCase()
+  if (id.includes("qwen3") || id.includes("qwen-3")) return 0.95
+  if (id.includes("qwen2.5-coder")) return 0.8
   if (id.includes("qwen")) return 1
   if (id.includes("gemini"))
     return GEMINI_MODELS_WITH_SAMPLING_DEFAULTS.some((model) => model.test(id)) ? 0.95 : undefined
diff --git a/packages/opencode/src/session/llm.ts b/packages/opencode/src/session/llm.ts
index a99f8acff20c..121a27e39674 100644
--- a/packages/opencode/src/session/llm.ts
+++ b/packages/opencode/src/session/llm.ts
@@ -109,6 +109,7 @@ const live: Layer.Layer<
         auth: info,
         plugin,
         flags,
+        samplingDefaults: cfg.experimental?.sampling_defaults,
         isWorkflow,
       })
 
diff --git a/packages/opencode/src/session/llm/request.ts b/packages/opencode/src/session/llm/request.ts
index 4f93411107df..e96a38c28a67 100644
--- a/packages/opencode/src/session/llm/request.ts
+++ b/packages/opencode/src/session/llm/request.ts
@@ -32,6 +32,7 @@ type PrepareInput = {
   readonly auth: Auth.Info | undefined
   readonly plugin: Plugin.Interface
   readonly flags: RuntimeFlags.Info
+  readonly samplingDefaults?: boolean
   readonly isWorkflow: boolean
 }
 
@@ -111,6 +112,18 @@ export const prepare = Effect.fn("LLMRequestPrep.prepare")(function* (input: Pre
           ...input.messages,
         ]
 
+  const auto = (input.samplingDefaults ?? true)
+    ? {
+        temperature: ProviderTransform.temperature(input.model),
+        topP: ProviderTransform.topP(input.model),
+        topK: ProviderTransform.topK(input.model),
+      }
+    : {
+        temperature: undefined,
+        topP: undefined,
+        topK: undefined,
+      }
+
   const params = yield* input.plugin.trigger(
     "chat.params",
     {
@@ -121,11 +134,9 @@ export const prepare = Effect.fn("LLMRequestPrep.prepare")(function* (input: Pre
       message: input.user,
     },
     {
-      temperature: input.model.capabilities.temperature
-        ? (input.agent.temperature ?? ProviderTransform.temperature(input.model))
-        : undefined,
-      topP: input.agent.topP ?? ProviderTransform.topP(input.model),
-      topK: ProviderTransform.topK(input.model),
+      temperature: input.model.capabilities.temperature ? (input.agent.temperature ?? auto.temperature) : undefined,
+      topP: input.agent.topP ?? auto.topP,
+      topK: auto.topK,
       maxOutputTokens: ProviderTransform.maxOutputTokens(input.model, input.flags.outputTokenMax),
       options,
     },
diff --git a/packages/opencode/test/provider/transform.test.ts b/packages/opencode/test/provider/transform.test.ts
index 701658987402..df3ab13f9419 100644
--- a/packages/opencode/test/provider/transform.test.ts
+++ b/packages/opencode/test/provider/transform.test.ts
@@ -3215,6 +3215,51 @@ describe("ProviderTransform.temperature - Cohere North", () => {
   })
 })
 
+describe("ProviderTransform sampling defaults - Qwen", () => {
+  const model = (id: string) =>
+    ({
+      id: `alibaba/${id}`,
+      api: { id },
+    }) as any
+
+  test.each([
+    "qwen3-coder-next",
+    "qwen3.8",
+    "qwen3.8-max",
+  ])("uses 0.95 top_p and 1.0 temperature for %s", (id) => {
+    expect(ProviderTransform.topP(model(id))).toBe(0.95)
+    expect(ProviderTransform.temperature(model(id))).toBe(1.0)
+  })
+
+  test.each([
+    "qwen3",
+    "qwen-3",
+    "qwen3-max",
+    "qwen3.5-plus",
+    "qwen3.6-plus",
+    "qwen3.7-max",
+  ])("uses 0.95 top_p and 0.6 temperature for %s", (id) => {
+    expect(ProviderTransform.topP(model(id))).toBe(0.95)
+    expect(ProviderTransform.temperature(model(id))).toBe(0.6)
+  })
+
+  test.each([
+    "qwen2.5-coder",
+    "qwen2.5-coder-32b-instruct",
+  ])("uses 0.8 top_p and 0.7 temperature for %s", (id) => {
+    expect(ProviderTransform.topP(model(id))).toBe(0.8)
+    expect(ProviderTransform.temperature(model(id))).toBe(0.7)
+  })
+
+  test.each([
+    "qwen-plus",
+    "qwen-max",
+  ])("keeps 1 top_p and 0.55 temperature for %s", (id) => {
+    expect(ProviderTransform.topP(model(id))).toBe(1)
+    expect(ProviderTransform.temperature(model(id))).toBe(0.55)
+  })
+})
+
 describe("ProviderTransform sampling defaults - Gemini", () => {
   const model = (id: string) =>
     ({
diff --git a/packages/sdk/js/src/v2/gen/types.gen.ts b/packages/sdk/js/src/v2/gen/types.gen.ts
index 90c91e9158cc..8fda9f84cce4 100644
--- a/packages/sdk/js/src/v2/gen/types.gen.ts
+++ b/packages/sdk/js/src/v2/gen/types.gen.ts
@@ -2025,6 +2025,7 @@ export type Config = {
     primary_tools?: Array<string>
     continue_loop_on_deny?: boolean
     mcp_timeout?: number
+    sampling_defaults?: boolean
     policies?: Array<ConfigV2ExperimentalPolicy>
   }
 }
diff --git a/packages/sdk/openapi.json b/packages/sdk/openapi.json
index afe14bb604cc..556e1a21d32d 100644
--- a/packages/sdk/openapi.json
+++ b/packages/sdk/openapi.json
@@ -21543,6 +21543,9 @@
                 "type": "integer",
                 "exclusiveMinimum": 0
               },
+              "sampling_defaults": {
+                "type": "boolean"
+              },
               "policies": {
                 "type": "array",
                 "items": {