From 7c0889674d16e607f1522dc833104626ab494cb4 Mon Sep 17 00:00:00 2001
From: ionianFury <09_wintry.carhops@icloud.com>
Date: Thu, 21 May 2026 15:36:26 +0200
Subject: [PATCH] design: bump per-image OpenAI timeout from 120s to 240s
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The hardcoded 120-second abort in design/src/{generate,iterate,variants,evolve}.ts
fires too often against the gpt-image-1 image_generation tool at the
default quality:"high" + size:"1536x1024". A real session today had
all 3 `variants` calls timeout at exactly 120s while the OpenAI API
was actively processing — direct curl with max-time 240s returned 200
in 33s.

Bumping each call to 240s gives ~2x headroom over typical worst-case
(60-180s observed) while staying under user patience. Quality default
unchanged — high is what produces the polished mockups users expect;
the timeout was the cliff, not the quality.

Touches 5 sites:
- design/src/generate.ts:40
- design/src/iterate.ts:85, 133
- design/src/variants.ts:61
- design/src/evolve.ts:55

All carry an inline comment explaining the picked value. A future
follow-up could promote this to an env var (GSTACK_DESIGN_TIMEOUT_MS)
or a shared constant in design/src/constants.ts, but I'm keeping this
PR small.

Verified: bun install + bun build --compile design/src/cli.ts succeeds.
---
 design/src/evolve.ts   | 3 ++-
 design/src/generate.ts | 5 ++++-
 design/src/iterate.ts  | 6 ++++--
 design/src/variants.ts | 3 ++-
 4 files changed, 12 insertions(+), 5 deletions(-)

diff --git a/design/src/evolve.ts b/design/src/evolve.ts
index c88ae6c660..f23b759445 100644
--- a/design/src/evolve.ts
+++ b/design/src/evolve.ts
@@ -52,7 +52,8 @@ export async function evolve(options: EvolveOptions): Promise<void> {
   ].join("\n");
 
   const controller = new AbortController();
-  const timeout = setTimeout(() => controller.abort(), 120_000);
+  // 240s — matches generate.ts; see comment there for rationale.
+  const timeout = setTimeout(() => controller.abort(), 240_000);
 
   try {
     const response = await fetch("https://api.openai.com/v1/responses", {
diff --git a/design/src/generate.ts b/design/src/generate.ts
index 383c51aeeb..fa18cc5220 100644
--- a/design/src/generate.ts
+++ b/design/src/generate.ts
@@ -37,7 +37,10 @@ async function callImageGeneration(
   quality: string,
 ): Promise<{ responseId: string; imageData: string }> {
   const controller = new AbortController();
-  const timeout = setTimeout(() => controller.abort(), 120_000);
+  // 240s: gpt-image-1 at quality:"high" size:"1536x1024" routinely takes
+  // 60-180s and can spike past 120s under provider load (observed in
+  // production). 240s gives ~2x headroom while staying under user patience.
+  const timeout = setTimeout(() => controller.abort(), 240_000);
 
   try {
     const response = await fetch("https://api.openai.com/v1/responses", {
diff --git a/design/src/iterate.ts b/design/src/iterate.ts
index c85eacee9b..bca3f7c752 100644
--- a/design/src/iterate.ts
+++ b/design/src/iterate.ts
@@ -82,7 +82,8 @@ async function callWithThreading(
   feedback: string,
 ): Promise<{ responseId: string; imageData: string }> {
   const controller = new AbortController();
-  const timeout = setTimeout(() => controller.abort(), 120_000);
+  // 240s — matches generate.ts; see comment there for rationale.
+  const timeout = setTimeout(() => controller.abort(), 240_000);
 
   try {
     const response = await fetch("https://api.openai.com/v1/responses", {
@@ -130,7 +131,8 @@ async function callFresh(
   prompt: string,
 ): Promise<{ responseId: string; imageData: string }> {
   const controller = new AbortController();
-  const timeout = setTimeout(() => controller.abort(), 120_000);
+  // 240s — matches generate.ts; see comment there for rationale.
+  const timeout = setTimeout(() => controller.abort(), 240_000);
 
   try {
     const response = await fetch("https://api.openai.com/v1/responses", {
diff --git a/design/src/variants.ts b/design/src/variants.ts
index d52eb22829..c5c7df27e4 100644
--- a/design/src/variants.ts
+++ b/design/src/variants.ts
@@ -58,7 +58,8 @@ export async function generateVariant(
     skipLeadingDelay = false;
 
     const controller = new AbortController();
-    const timeout = setTimeout(() => controller.abort(), 120_000);
+    // 240s — matches generate.ts; see comment there for rationale.
+    const timeout = setTimeout(() => controller.abort(), 240_000);
 
     try {
       const response = await fetchFn("https://api.openai.com/v1/responses", {