"Twitch is down" almost never means all of Twitch is down. Live video delivery, chat, login, the creator dashboard, clips, and drops run on separate backends that fail independently, so an infinite purple loading spinner and a "you were logged out" prompt are different incidents with different fixes. Before you clear caches or restart your router, work out which component broke and whether anyone else is seeing it.
The parts of Twitch that fail separately ¶
- Video playback / CDN. The live video is served from a delivery layer separate from the website. Streams can refuse to load while the site, the browse directory, and chat all work normally.
- Chat. Chat rides its own connection. During video incidents it is common to see chat scrolling under a dead player, and occasionally the reverse, a playing stream with a chat that will not connect.
- Login / auth. Sign-in, two-factor prompts, session validation. An auth incident logs people out or blocks sign-in while anonymous viewing keeps working.
- Creator dashboard / Studio. The tools streamers use to go live and manage a channel. The dashboard can be broken while every live stream keeps playing, so viewers notice nothing.
- Clips and VODs. Recorded content sits on separate storage and processing. Clips failing to load, or a VOD stuck processing, says nothing about live video.
- Drops and inventory. Drops progress tracking stalls independently of everything else, and delayed drops are one of the most misreported "Twitch is down" symptoms. Progress usually reconciles later.
Symptom to component ¶
| What you see | Likely component | Likely scope |
|---|---|---|
| Error 2000 (network error) | Player, in your browser | Usually local, an extension or security software interfering with the player |
| Error 3000 (media decoding error) | Player / browser | Usually local, browser cache or decode problem |
| Error 4000 (format not supported) | Player / browser | Usually local, another app or driver holding the media pipeline |
| Purple spinner forever, on every channel | Video delivery | Platform-side if other devices and networks see it too |
| Chat works, video does not | Video delivery | Platform-side when many channels are affected at once |
| Video plays, chat will not connect | Chat backend | Platform-side, unless it is one channel's chat settings |
| Logged out everywhere, sign-in fails | Auth | Platform-side, wait instead of retrying passwords |
| Dashboard errors while streams play fine | Dashboard / Studio | Platform-side for creators, viewers unaffected |
The pattern worth memorizing: the numbered 2000-series, 3000-series, and 4000-series errors are player-level and usually local, while the same failure on every channel at once is platform-level or network-level.
Check the scope in under a minute ¶
- Probe twitch.tv with 4 methods. Run a check on twitch.tv. It sends 4 independent probe methods in parallel, HTTP/1.1 with strict TLS, HTTP/2 with relaxed TLS, a browser-style GET, and a curl-style HEAD. All four failing means the site itself is unreachable, a real outage. A mixed result means the site is up but blocking or mishandling one request type. All four passing while your stream will not play means the website is fine and the problem is in the video path or your browser, keep going.
- Read Twitch's own status feed. The Twitch publisher page polls Twitch's official status feed every 5 minutes and shows the overall indicator plus any open incidents by name. Twitch names the affected component in the incident title, video, chat, login, and the rest, so if the feed lists an open incident on video viewing, stop troubleshooting your browser.
- The one-streamer test. Open two or three large channels from the front page. If only one channel is down, that is the streamer, their internet, their encoder, or their power, not Twitch. If every channel fails the same way, it is Twitch or your side, and step 1 splits those two.
This is the same triage as the general down-for-everyone-or-just-me diagnostic, specialized for a platform where the components fail separately.
If it is you: local fixes, honestly framed ¶
These steps fix player-level problems, the error 2000/3000/4000 family. They do nothing for a platform incident, so check the status feed first and skip this section if Twitch has already confirmed one.
- Disable extensions, especially ad blockers. The single most common cause of error 2000. Ad blockers and privacy extensions intercept the player's network requests, and a filter-list update can break playback overnight with no change on your end. Test in an incognito window with extensions off before uninstalling anything.
- Try a different browser. If the stream plays in another browser, the problem is your primary browser's cache, extensions, or hardware-acceleration settings, not Twitch and not your network.
- Cross-check on the mobile app, on cellular. Take Wi-Fi out of the equation. If the app plays the stream your desktop cannot, the fault is in your desktop browser or home network. If both fail, the evidence points back at Twitch.
- Clear site data for twitch.tv. Error 3000 in particular tends to follow a corrupted cache. Clearing cookies and cached files for the site alone is enough, you do not need to wipe the whole browser.
Be honest with yourself about the order here. If the one-streamer test showed every channel failing and the probe check failed too, none of these steps will help, and the cross-check on cellular is the only one worth doing.
When to just wait ¶
If the publisher feed shows an active incident on the component you are hitting, waiting is the correct move. There is no local workaround for a video delivery incident, and hammering the login form during an auth incident only risks tripping rate limits on your account. Streamers mid-broadcast during an ingest or dashboard incident should generally keep the encoder running rather than restarting it, reconnects usually resume cleanly once the incident clears.
If you rely on Twitch for work, monitoring beats refreshing. A free isitdown.io account can watch twitch.tv with email alerts. Incidents open after 2 consecutive failed 5-minute probes and resolve after 2 consecutive passes, so an alert means a sustained failure, not a single blip. The same component-first triage applies to other real-time platforms, see the Discord guide for the equivalent walkthrough there.
FAQ ¶
Is Twitch down or is it just me?
Run the three checks in order: probe twitch.tv with all four methods, read the Twitch status feed, and open several unrelated channels. All probes failing means a real outage. Probes passing while every channel's video fails on multiple devices points at a video delivery incident. One dead channel among working ones is the streamer, not the platform.
What does Twitch error 2000 mean?
Error 2000 is a network error raised by the player, and in practice it is usually caused by something in your browser, an ad blocker, a privacy extension, or security software interfering with the player's connection. Test in an incognito window with extensions disabled. If the error persists across browsers and devices on the same channel set, treat it as a platform problem instead.
Why does chat work when the video will not load?
Because they are different systems. Chat uses its own connection to a separate backend, while video comes from the delivery network. A scrolling chat under a dead player means your connection to Twitch is fine and the video path specifically is failing, which is a platform-side signal when it affects many channels at once.
My drops are not progressing, is Twitch down?
Almost certainly not. Drops tracking runs separately from video and chat and it lags or stalls on its own. If streams play and chat connects, the platform is up. Drops progress typically reconciles after a delay, so leave the stream running and check your inventory later rather than refreshing.