Claude Code OAuth login fails with a timeout error. `auth.anthropic.com` does not resolve via DNS, making it impossible to authenticate.

Found 3 possible duplicate issues:

1. https://github.com/anthropics/claude-code/issues/33213
2. https://github.com/anthropics/claude-code/issues/33214
3. https://github.com/anthropics/claude-code/issues/33217

This issue will be automatically closed as a duplicate in 3 days.

• If your issue is a duplicate, please close it and 👍 the existing issue instead
• To prevent auto-closure, add a comment or 👎 this comment

🤖 Generated with Claude Code

Experiencing the same issue on Windows 10, Claude Code version 2.1.72. The final step of the login flow times out waiting for the server response. Tried approximately 10 times without success. The browser and email steps complete fine but the server confirmation exceeds the timeout window.

same issue. Ubuntu 24.04.4 LTS
location California
i can log in through the web browser but not claude code CLI. /login command fails.

+1

me too... here korea

Workaround for Claude Cli timeout issues on Windows (I am able to login now)

What I did:

Located the global npm install folder: C:\Users\<yourname>\AppData\Roaming\npm\node_modules\@anthropic-ai\claude-code\

The main file is cli.js (it's minified so it's one giant unreadable line). Find and replace the hardcoded timeout values:

timeout:15000 → timeout:45000

AbortSignal.timeout(15000) → AbortSignal.timeout(45000)


That's it. Just save the file and restart claude cli. I am able to /login and authenticate successfully

same here on mac, cant login again

The fix above works for mac too. See here

https://github.com/anthropics/claude-code/issues/33239#issuecomment-4040282274

### Workaround for Claude Cli timeout issues on Windows (I am able to login now) What I did: Located the global npm install folder: C:\Users<yourname>\AppData\Roaming\npm\node_modules@anthropic-ai\claude-code\ The main file is cli.js (it's minified so it's one giant unreadable line). Find and replace the hardcoded timeout values: timeout:15000 → timeout:45000 AbortSignal.timeout(15000) → AbortSignal.timeout(45000) That's it. Just save the file and restart claude cli. I am able to /login and authenticate successfully

This workaround only applies if Claude Code was installed via npm, which is now considered a deprecated installation method.

this works for me

$file = "C:\nvm4w\nodejs\node_modules\@anthropic-ai\claude-code\cli.js"
$content = Get-Content $file -Raw
$content = $content -replace 'timeout:15000', 'timeout:45000'
$content = $content -replace 'AbortSignal\.timeout\(15000\)', 'AbortSignal.timeout(45000)'
Set-Content $file -Value $content -NoNewline

Same issue on EndeavourOS (Arch-based), Lima, Peru. Claude Code 2.1.72, installed via npm + nvm.

Environment:

• OS: EndeavourOS / kernel 6.19.6-arch1-1
• Node: v22.14.0 (nvm)
• Shell: bash 5.3.9
• Can log in via claude.ai web browser ✅
• /login in CLI fails with OAuth error: timeout of 15000ms exceeded ❌

Workaround confirmed working for Linux/nvm users:

# Find the file
CLI=$(find ~/.nvm/versions/node -name 'cli.js' -path '*claude-code*')
echo "Patching: $CLI"

# Apply the timeout fix
sed -i 's/timeout:15000/timeout:45000/g' "$CLI"
sed -i 's/AbortSignal.timeout(15000)/AbortSignal.timeout(45000)/g' "$CLI"

After patching, /login completes successfully. This confirms the issue is not specific to Windows — it affects Linux users in Latin America as well, likely due to regional latency on the OAuth callback endpoint.

does anyone have ip for auth.anthropic.com ?

I changed the version to stable claude install stable and managed to log in.

I have a similar issue where when logging in via CLI in the vscode, the URL will open in the default browser. After pasting the code to the CLI, it returns Login OAuth error: timeout of 15000ms exceeded.
I am located in Southern California.

The issue was fixed after rolling back from 2.1.72 to a previous version.

Confirming the same issue from Bangkok, Thailand

I'm experiencing identical behavior:

Environment:

• Claude Code version: 2.1.72
• Operating System: Linux (Ubuntu)
• Location: Bangkok, Thailand
• Network: Home network (no VPN, no proxy)

DNS Resolution:

$ nslookup auth.anthropic.com
** server can't find auth.anthropic.com: NXDOMAIN

$ nslookup auth.anthropic.com 8.8.8.8
** server can't find auth.anthropic.com: NXDOMAIN

$ nslookup api.anthropic.com
Name:   api.anthropic.com
Address: 160.79.104.10

Same pattern - api.anthropic.com resolves fine, but auth.anthropic.com returns NXDOMAIN from both local DNS and Google DNS (8.8.8.8).

This appears to be affecting Southeast Asia region broadly (Malaysia + Thailand confirmed). OAuth login is impossible without VPN or API key workaround.

Same here on windows 10. It says "build something great", but the console has already timed out

I changed the version to stable claude install stable and managed to log in.

yep. that works for me

Fix: Claude Code CLI Timeout

Problem

Claude Code CLI has a default timeout of 15 seconds (15000ms) which can cause disconnections or failures on slow connections.

Solution

Increase the timeout to 45 seconds (45000ms) by editing the cli.js file directly.

---

macOS

1. Find where Claude is installed

which claude

This will return something like /opt/homebrew/bin/claude.

To get the cli.js path, replace bin/claude with lib/node_modules/@anthropic-ai/claude-code/cli.js.

For example, if which claude returns /opt/homebrew/bin/claude:

CLI_PATH="/opt/homebrew/lib/node_modules/@anthropic-ai/claude-code/cli.js"

2. Replace the timeouts

sed -i '' 's/timeout:15000/timeout:45000/g' "$CLI_PATH"
sed -i '' 's/AbortSignal.timeout(15000)/AbortSignal.timeout(45000)/g' "$CLI_PATH"

3. Verify

grep "45000" "$CLI_PATH"

---

Linux

1. Find where Claude is installed

which claude

Same logic: replace bin/claude with lib/node_modules/@anthropic-ai/claude-code/cli.js.

2. Replace the timeouts

Note: On Linux, sed -i goes without ''.

sed -i 's/timeout:15000/timeout:45000/g' "$CLI_PATH"
sed -i 's/AbortSignal.timeout(15000)/AbortSignal.timeout(45000)/g' "$CLI_PATH"

3. Verify

grep "45000" "$CLI_PATH"

---

Windows

1. Find where Claude is installed

where claude

This will return something like C:\Users\YourUser\AppData\Roaming\npm\claude.

The cli.js path will be under node_modules\@anthropic-ai\claude-code\cli.js in the same root.

For example:

$CLI_PATH = "C:\Users\YourUser\AppData\Roaming\npm\node_modules\@anthropic-ai\claude-code\cli.js"

2. Replace the timeouts

(Get-Content $CLI_PATH) -replace 'timeout:15000', 'timeout:45000' -replace 'AbortSignal.timeout\(15000\)', 'AbortSignal.timeout(45000)' | Set-Content $CLI_PATH

3. Verify

Select-String -Path $CLI_PATH -Pattern "45000"

---

Note

This change is lost on update. You'll need to repeat it after each Claude Code update.

I'm not sure why increasing the timeout seems to work for some people, but that doesnt work for me because I still can't resolve the auth.anthropic.com hostname, here is how I fixed it on Windows:

Run Notepad as Administrator -> Open: C:\Windows\System32\drivers\etc\hosts

Add these lines to the bottom of the file and save it:

160.79.104.1   auth.anthropic.com
160.79.104.2   auth.anthropic.com
160.79.104.3   auth.anthropic.com

Open a Terminal session as Admin and run:

ipconfig /flushdns

Profit.

no login in istanbul unfortunately by today

Same problem. Ubuntu. I can't tweak the timeout since I have a binary dist of claude installed the official way. The DNS workaround above by HauteMic does not work for me. On the browser, after clicking "Authorize" I can see "Waiting for localhost" on the status bar.

Edit: after trying 20 times it finally worked.

same problem here in Singapore

Resolved on Ubuntu 20.04 by installing the stable version

I was on the latest version (2.1.83) and encountered the same OAuth error: timeout of 15000ms exceeded issue.

Installing the stable version fixed it:

curl -fsSL https://claude.ai/install.sh | bash -s stable

After installation, authentication completed successfully.

📎 Install guide: https://code.claude.com/docs/en/setup

Today 6th of April, same problem in Barcelona, Spain

Today 6th of April, same problem in Istanbul, Turkey

Today 6th of April, same problem in Barcelona, Spain

Experiencing the same now in South Africa

Experiencing the same now in US

Today 6th of April, same problem in San Francisco, CA, United States

Today 6th of April, same problem in Istanbul, Turkey

London too

Same on Brazil

Germany too

Same in Sweden

Experiencing same issue here in Nairobi, Kenya

Also in the United States

this feels like a group chat haha. love it. we're suffering partners here

Also in the Iligan City, Philippines

Same in Taipei.

Happened on me also, from Indonesia.

OAuth error: timeout of 15000ms exceeded

nslookup auth.anthropic.com 1.1.1.1
Server:		1.1.1.1
Address:	1.1.1.1#53

** server can't find auth.anthropic.com: NXDOMAIN

same here 🙂 i 'm so glad i found this thread

Same issue now from argentina

Reykjavik, Iceland as well.

Obviously as of today this has escalated from a local issue to more widespread

Same here on Mars

OAuth error: Request failed with status code 500

Singapore, same here

same here in the Arthemis rocket

what now anthropic?

maybe they just leaked their source code again

Come on man, I need to provide stakeholder value on this Monday

<img width="1388" height="1033" alt="Image" src="https://github.com/user-attachments/assets/8e51cc8f-97c7-4c87-a149-71eabab905f2" />

It's always DNS:

https://dnschecker.org/#A/auth.anthropic.com

Guess this issue arond the world! Claude is down

Same here - Canada!

Serbia as well

It's also down here. Any solution?

Same issue - US.
It was working fine yesterday.
I guess I have to actually google stuff and code manually now... just like our ancestors did pre-2025.

Same issue in Argentina

Same here, Western Australia. Anyone up for some CS2 Arms Race while we wait? 😆

Stretch them fingers out boys

ah really, fails for me too haha

Same

Same here, Western Australia. Anyone up for some CS2 while we wait? 😆

add me 1527955959

Same here in Brazil

Same here in the US

same here in California

Austria... down

glad to know I'm in good company. On a mac in eastern US.

same for me too

Same here in New Delhi, but has anyone found a workaround for this ?
when its not installed via npm ? on mac.

same here. macOS, latest claude code version. germany

the same

<img width="1045" height="366" alt="Image" src="https://github.com/user-attachments/assets/30e87850-a3f6-4854-bf50-2e5a43543b97" />

Sorry, DNS wasn't a part of our deal

Same here in Switzerland

Same issue - US. It was working fine yesterday. I guess I have to actually google stuff and code manually now... just like our ancestors did pre-2025.

Just wait for fix and do nothing. It has to be free day while Claude is down.

<img width="1293" height="511" alt="Image" src="https://github.com/user-attachments/assets/56d7732c-ad04-4364-b157-675444a1d4d6" />

The Claude app works, i can still use claude code there. maybe because i am logged in in the first place.

To think we all are relying on this single tool this much.
But has anyone's work has been actually affected ? like, where you cant even make any progress ?

Same here

But has anyone's work has been actually affected ? like, where you cant even make any progress ?

I use Cursor now 😆

To think we all are relying on this single tool this much. But has anyone's work has been actually affected ? like, where you cant even make any progress ?

I'd just finished a big plan and was about to execute it when my login stopped working 😩

Same issue here, Buenos Aires Argentina.

same issue

Same issue in Czech Republic

same issue LOL

Same issue :D
Tried to code another BTC but my brain said “nah bro, that’s tomorrow’s problem”

I patched the binary to increase the timeout threshold and it worked:
Linux sed -i 's/timeout:15000/timeout:45000/g' "$(readlink -f $(which claude))"

Same issue in Israel

for the mac homebrew users feeling _frisky_ and want to patch their bins, you will need to self sign it afterwards.

#!/bin/bash

VERSION="2.1.92"
CLAUDE_BIN="$HOME/.local/share/claude/versions/$VERSION"

# Back up
cp "$CLAUDE_BIN" "$CLAUDE_BIN.bak"

# Patch (perl handles binary files better than sed on macOS)
perl -i -pe 's/timeout:15000/timeout:45000/g; s/AbortSignal\.timeout\(15000\)/AbortSignal.timeout(45000)/g' \
    "$CLAUDE_BIN"

# Re-sign (required on macOS after binary modification)
codesign --force --sign - "$CLAUDE_BIN"

tested this with my personal max account and my work enterprise

I patched the binary to increase the timeout threshold and it worked: Linux sed -i 's/timeout:15000/timeout:45000/g' "$(readlink -f $(which claude))"

Wow this working, thanks

anyone tried with brew installation for macos?

On WSL increasing the timeout threshold works!

sed -i 's/timeout:15000/timeout:45000/g' "$(readlink -f $(which claude))"

Skribble waiting room is fire guys

Even I'm facing the same issue. Tried logging in from vscode and also through terminal using claude CLI but not able to use and it gives same error of "OAuth error: timeout of 15000ms exceeded".
Did anyone found a solution or workaround for this issue?

Same here, on macOS NXDomain for auth.anthropic.com

Same problem

same here got this  Please run /login · API Error: 401 {"type":"error","error":{"type":"authentication_error","message":"OAuth token has expired. Please obtain a new token or refresh your existing
token."},"request_id":"req_011CZnoK1pScAQXT1CebbcjR"}

same problem on macos

On macOS, pasting a shebang line (#!/bin/bash) directly into zsh causes event not found - zsh treats ! as a history expansion character.

Fix: wrap your heredoc delimiter in single quotes (<< 'EOF') to disable all special character interpretation.

TLDR. Macos solution

```cat > patch_claude.sh << 'EOF'
#!/bin/bash
VERSION="2.1.92"
CLAUDE_BIN="$HOME/.local/share/claude/versions/$VERSION"
cp "$CLAUDE_BIN" "$CLAUDE_BIN.bak"
perl -i -pe 's/timeout:15000/timeout:45000/g; s/AbortSignal\.timeout\(15000\)/AbortSignal.timeout(45000)/g' \
"$CLAUDE_BIN"
codesign --force --sign - "$CLAUDE_BIN"
EOF
chmod +x patch_claude.sh
./patch_claude.sh

universal for macos, works for brew?

#!/bin/bash

CLAUDE_BIN=$(which claude)

# Back up
cp "$CLAUDE_BIN" "$CLAUDE_BIN.bak"

# Patch (perl handles binary files better than sed on macOS)
perl -i -pe 's/timeout:15000/timeout:45000/g; s/AbortSignal\.timeout\(15000\)/AbortSignal.timeout(45000)/g' \
    "$CLAUDE_BIN"

# Re-sign (required on macOS after binary modification)
codesign --force --sign - "$CLAUDE_BIN"

but it didnt help at all, now getting error 500 instead of timeout
OAuth error: Request failed with status code 500

for the mac homebrew users feeling _frisky_ and want to patch their bins, you will need to self sign it afterwards. #!/bin/bash VERSION="2.1.92" CLAUDE_BIN="$HOME/.local/share/claude/versions/$VERSION" # Back up cp "$CLAUDE_BIN" "$CLAUDE_BIN.bak" # Patch (perl handles binary files better than sed on macOS) perl -i -pe 's/timeout:15000/timeout:45000/g; s/AbortSignal\.timeout\(15000\)/AbortSignal.timeout(45000)/g' \ "$CLAUDE_BIN" # Re-sign (required on macOS after binary modification) codesign --force --sign - "$CLAUDE_BIN" tested this with my personal max account and my work enterprise

it worked with me too, thanks a lot

anyone else getting OAuth error: Request failed with status code 500 after patching timeout for macos?

any workaround for windows native installation?

Update - This issue additionally affects login on other surfaces, such as Claude.code.
Apr 06, 2026 - 15:54 UTC

https://status.claude.com/

Fix your sht Anthropic!

same here, on MacOs, Colombia

Anyone got a solution for Nobara Linux/Fedora?

Can't login too; MacOS, Indonesia

same here, on Mac-Os

Same!!!

macOS Paraguay, same

@lovetrol what you mentioned cannot be done as anthropic discontinued npm package.

Okay so the error seems to be this one now, on mac

<img width="633" height="272" alt="Image" src="https://github.com/user-attachments/assets/937c17cf-9b91-4fb0-8002-11a1480f46e3" />

i am also facing the same issue .. not sure why did it log out

Okay so the error seems to be this one now, on mac <img alt="Image" width="633" height="272" src="https://private-user-images.githubusercontent.com/46109434/574231324-937c17cf-9b91-4fb0-8002-11a1480f46e3.png?jwt=eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.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.2oMqgBnUG5Zm340hEfHO-qTtHRJwbLo8qKiqQVsvfYM">

The status page now acknowledges the issue. so I assume it's related to them working on it

+1

> Okay so the error seems to be this one now, on mac > <img alt="Image" width="633" height="272" src="https://private-user-images.githubusercontent.com/46109434/574231324-937c17cf-9b91-4fb0-8002-11a1480f46e3.png?jwt=eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.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.2oMqgBnUG5Zm340hEfHO-qTtHRJwbLo8qKiqQVsvfYM"> The status page now acknowledges the issue. so I assume it's related to them working on it

i hope so

Same here. NixOS, Korea, v2.1.81, home network (no VPN)

+1

Hi friends! Sorry for the trouble and thank you for your reports. We're are working on this issue now. You can follow along here

<img width="874" height="254" alt="Image" src="https://github.com/user-attachments/assets/156a9e71-3d68-409c-bee4-5c565d51e9e5" />
is back

hey can't login into claude code right now

Removing cached credentials and logging in with /login worked for me.

NOTE: Possible false positive - the service may be recovering on its own.

# macOS only
# security delete-generic-password -s "claude-code" 2>/dev/null

# Back up cached credentials (optional)
cp ~/.claude/.credentials.json ./.credentials-backup.json

# Remove cached credentials
rm -f ~/.claude/.credentials.json

# Start claude code. NOTE: `claude login` may fail. Use `/login` from within a session.
claude

/login

working in sf
you can stop touching the grass now

I think the issue is fixed (US central). I can login now. Didn't change anything. Looks like anthropic's auth got fixed.

<img width="1148" height="128" alt="Image" src="https://github.com/user-attachments/assets/2ae60d26-7daf-4046-8d19-f4c3a628cc99" />

Now it is working right

<img width="211" height="51" alt="Image" src="https://github.com/user-attachments/assets/b8465872-1ed4-4e8b-b071-541601e9ff31" />

It's back! Thanks guys!

Issue is back for me

still the same it is not working tho

I am getting 500 even

Getting 500 or OAuth error: timeout of 15000ms exceeded

It's being more than one week that this problem is going on... Now again

same issue as yesterday, OAuth error: timeout of 15000ms exceeded

I get the same issue last few hours

OAuth error: Request failed with status code 500

i am in macos btw

it seems server-side

hope they are fixing it rn

same, from timeout to:

<img width="358" height="31" alt="Image" src="https://github.com/user-attachments/assets/a3c73a3d-44e5-44bc-bea7-eabb6957719d" />

Edit: Claude Desktop (Windows) seems to be unaffected, still works with previously logged in credentials

OAuth error: Request failed with status code 500

Same here, either code 500 or OAuth error: timeout of 15000ms exceeded when i use the link. I want to use the remaining tokens in my account before my week limit resets in 2 hours!

Same here. login is toast.

Same here. Web says "all set up" but it doesn't work, it's either request failed with 500 or 1500ms exceeded

https://status.claude.com/incidents/124yr07585k9

This issue occurred yesterday and also today for me in USA region claude code for Windows v 2.1.92.
Please advise any fixes?

Getting this same issue any fix for windows?

Why the heck has anthropic hardcoded the timeout to 15 seconds. I even asked Claude if there was a way to increase the timeout if I installed Claude Code the recommended way on windows via powershell and it told me it's impossible.

Is Anthropic not willing to do anything about this? It's getting tiring having to put up with this issue all the time...

OAuth error: timeout of 15000ms exceeded

Same Error

I'm having issues too. Reinstalled claude code, latest version. Timeout, auth domain does not resolve.

Getting same issues here in the UK. reinstalled claude completely. Tried different browsers. Installed via official curl command not Brew. Using Macos.

❯ nslookup auth.anthropic.com 1.1.1.1
Server:		1.1.1.1
Address:	1.1.1.1#53

** server can't find auth.anthropic.com: NXDOMAIN

the auth domain doesn't resolve

update, everything working already here. try check it out guys

Apologies for the disruption, folks! We believe this issue has been resolved.

nslookup auth.anthropic.com 1.1.1.1
Server:         1.1.1.1
Address:        1.1.1.1#53

** server can't find auth.anthropic.com: NXDOMAIN

Das Problem besteht seit gestern unter Ubuntu in Deutschland

I'm still having this issue in Ubuntu WSL.