ChatGPT Network Error — Why It Happens on Long Responses and How to Fix It

ChatGPT streams the answer token by token over a single connection while it generates. Your ISP, a corporate firewall, a VPN, or an intermediate CDN hop will often kill a connection that stays open past roughly 60 seconds. A big code dump or a long essay can easily take longer than that to finish streaming — the connection gets cut mid-answer and the UI reports "network error," even though the model itself was generating correctly.

The most reliable fix is simply avoiding one giant response. Break the request up so no single answer needs to stream long enough to hit the timeout:

After a "continue," always sanity-check the boundary between the two chunks (e.g. a function that got cut mid-line) — ask ChatGPT to re-show just that section if anything looks incomplete.

Ad-blockers, privacy extensions, and third-party "ChatGPT enhancer" or prompt-manager extensions can intercept or interrupt the streaming connection. A stale session or corrupted local cache causes similar symptoms:

Unstable Wi-Fi, VPNs, and corporate proxies are especially prone to killing long-lived streaming connections before they finish:

Occasionally the cause really is on OpenAI's side. Check prismix.dev/service/openai for an active incident before assuming it's your setup — if status is normal and only long responses fail, it's the streaming timeout described above, not an outage.

If you're generating long output programmatically rather than in the chat UI, the API gives you far more control than the web app — you own the timeout and can reconnect or resume on your own terms:

// Node — stream with your own generous timeout instead of
// relying on the browser tab's connection lifetime
import OpenAI from 'openai';

const client = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });

const controller = new AbortController();
const timeout = setTimeout(() => controller.abort(), 300_000); // 5 min

const stream = await client.chat.completions.create(
  {
    model: 'gpt-4o',
    messages: [{ role: 'user', content: 'Write the full module...' }],
    stream: true,
  },
  { signal: controller.signal }
);

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? '');
}
clearTimeout(timeout);

See OpenAI API not working for broader API-side troubleshooting.