From 7cdecd824a871efce161fe68aa9e230ca8a8788b Mon Sep 17 00:00:00 2001
From: roboomp <omp@can.ac>
Date: Tue, 14 Jul 2026 15:56:49 +0000
Subject: [PATCH] fix(catalog): widened GLM coding-plan idle timeout to
 opencode gateways

GLM-5.x coding-plan SKUs idle for minutes mid-reasoning, so they get a
600s stream idle-timeout floor instead of the 120s default. That floor
was gated to the native Z.AI/Zhipu hosts only, so GLM-5.2 served through
the OpenCode Go/Zen gateways fell back to the 120s watchdog and stalled
with "OpenAI completions stream stalled while waiting for the next event"
during the slow /plan writing phase.

Fixes #4758
---
 packages/catalog/CHANGELOG.md              |  4 +++
 packages/catalog/src/compat/openai.ts      |  2 +-
 packages/catalog/test/zhipu-compat.test.ts | 33 ++++++++++++++++++++++
 3 files changed, 38 insertions(+), 1 deletion(-)

diff --git a/packages/catalog/CHANGELOG.md b/packages/catalog/CHANGELOG.md
index 42e39218a..ee14f3ef7 100644
--- a/packages/catalog/CHANGELOG.md
+++ b/packages/catalog/CHANGELOG.md
@@ -2,6 +2,10 @@
 
 ## [Unreleased]
 
+### Fixed
+
+- Fixed GLM-5.x coding-plan streams via the OpenCode Go/Zen gateways (`opencode.ai/zen/…`) timing out with `OpenAI completions stream stalled while waiting for the next event` during slow plan-writing/reasoning phases. The 600s idle-timeout floor for GLM coding-plan SKUs was gated to the native Z.AI/Zhipu hosts only, so OpenCode-fronted GLM fell back to the 120s default watchdog. ([#4758](https://github.com/can1357/oh-my-pi/issues/4758))
+
 ## [16.3.11] - 2026-07-06
 
 ### Added
diff --git a/packages/catalog/src/compat/openai.ts b/packages/catalog/src/compat/openai.ts
index d5627ccd5..6f1dd0ed8 100644
--- a/packages/catalog/src/compat/openai.ts
+++ b/packages/catalog/src/compat/openai.ts
@@ -377,7 +377,7 @@ export function buildOpenAICompat(spec: ModelSpec<"openai-completions">): Resolv
 	// for minutes while reasoning or cold-loading weights; widen the idle
 	// timeout so warm-ups stop aborting and retrying.
 	const streamIdleTimeoutMs =
-		GLM_CODING_PLAN_MODEL_PATTERN.test(spec.id) && (isZai || isZhipu)
+		GLM_CODING_PLAN_MODEL_PATTERN.test(spec.id) && (isZai || isZhipu || isOpenCodeHost)
 			? GLM_CODING_PLAN_STREAM_IDLE_TIMEOUT_MS
 			: provider === "alibaba-coding-plan"
 				? ALIBABA_CODING_PLAN_STREAM_IDLE_TIMEOUT_MS
diff --git a/packages/catalog/test/zhipu-compat.test.ts b/packages/catalog/test/zhipu-compat.test.ts
index 82ff798be..b69b9afb7 100644
--- a/packages/catalog/test/zhipu-compat.test.ts
+++ b/packages/catalog/test/zhipu-compat.test.ts
@@ -120,6 +120,39 @@ describe("openai-completions compat — zhipu-coding-plan branch", () => {
 	});
 });
 
+describe("openai-completions compat — GLM coding-plan stream idle timeout", () => {
+	function glm52(provider: string, baseUrl: string): ModelSpec<"openai-completions"> {
+		return { ...baseModel, id: "glm-5.2", name: "GLM-5.2", provider, baseUrl };
+	}
+
+	// GLM coding-plan SKUs idle for minutes mid-reasoning; the 600s watchdog
+	// floor must apply on every gateway that fronts them, not just the native
+	// Z.AI/Zhipu hosts (issue #4758: GLM-5.2 via opencode-go stalled with
+	// "OpenAI completions stream stalled while waiting for the next event").
+	it("widens the idle timeout to 600s for GLM-5.x on Z.AI, Zhipu, and OpenCode gateways", () => {
+		expect(buildOpenAICompat(glm52("zai", "https://api.z.ai/api/coding/paas/v4")).streamIdleTimeoutMs).toBe(600_000);
+		expect(
+			buildOpenAICompat(glm52("zhipu-coding-plan", "https://open.bigmodel.cn/api/coding/paas/v4"))
+				.streamIdleTimeoutMs,
+		).toBe(600_000);
+		expect(buildOpenAICompat(glm52("opencode-go", "https://opencode.ai/zen/go/v1")).streamIdleTimeoutMs).toBe(
+			600_000,
+		);
+		expect(buildOpenAICompat(glm52("opencode-zen", "https://opencode.ai/zen/v1")).streamIdleTimeoutMs).toBe(600_000);
+	});
+
+	it("does not widen non-GLM models on the OpenCode gateway via the GLM floor", () => {
+		const kimi = buildOpenAICompat({
+			...baseModel,
+			id: "kimi-k2.5",
+			name: "Kimi K2.5",
+			provider: "opencode-go",
+			baseUrl: "https://opencode.ai/zen/go/v1",
+		});
+		expect(kimi.streamIdleTimeoutMs).toBeUndefined();
+	});
+});
+
 describe("zhipu-coding-plan model discovery", () => {
 	it("uses the dedicated Coding Plan endpoint by default", async () => {
 		let requestedUrl = "";