[BUG] API Error: Overloaded no longer automatically retrying
Resolved 💬 44 comments Opened Mar 31, 2025 by scottleibrand Closed Apr 3, 2025
💡 Likely answer: A maintainer (ashwin-ant, collaborator)
responded on this thread — see the highlighted reply below.
Environment
- Claude CLI version: 0.2.56 (Claude Code)
Bug Description
The retries for API errors added in 0.2.53 per #583 are no longer working in 0.2.56, as noted by @plucked and @santoshgistto in their comments on #583. I'm now seeing the same:
⎿ API Error: {"type":"error","error":{"details":null,"type":"overloaded_error","message":"Overloaded"} }Any chance you could re-add the retries from 0.2.53 please? If you had to remove it to reduce load, can you re-add it with longer timeouts? I don't really care how many seconds it takes to retry, as long as I don't have to intervene to get it to continue once load comes down.
44 Comments
I still have one session still running v0.2.55, and it still correctly retries API Errors:
So this is a regression/reversion in 0.2.56.
Workaround:
I successfully downgraded to 0.2.55 with that and automatic API retries are working again on the newly downgraded session as well.
We're investigating this, thanks for reporting!
happening for me too
@mtrazzi are you seeing this on the latest version (0.2.61)? We put in a fix to make retries work reliably again.
This is happening for me too
Happening right now.
happening for me too
!Image
just here to add to the choir!
thanks guys
me too
Our API backend is experiencing some issues right now: https://status.anthropic.com/incidents/s0c6w5gmlcq3
did anyone try this: npm i -g @anthropic-ai/claude-code@0.2.55 && claude does it work?
Same error for me
same error here
Things should be working now!
This is the retry feature working like it's supposed to:
The fact that the API backend was experiencing some issues per https://status.anthropic.com/incidents/s0c6w5gmlcq3 is a separate problem, not caused by this bug. This bug was about the client code not handling such API issues gracefully with retries, which now seems to be happening correctly in 0.2.61.
Thanks @ashwin-ant and everyone else for confirming that is fixed: now I can upgrade from 0.2.55. :-)
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 1 seconds… (attempt 1/10)
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 1 seconds… (attempt 2/10)
⎿ We are experiencing high demand for Opus 4.
Seeing this for the first time today.
issue still persist (Mac OS, Claude Code through VsCode Terminal, )
✶ Organizing… (161s · ↑ 0 tokens
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 1 seconds… (attempt 1/10)
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 1 seconds… (attempt 2/10)
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 2 seconds… (attempt 3/10)
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 5 seconds… (attempt 4/10)
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 10 seconds… (attempt 5/10)
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 18 seconds… (attempt 6/10)
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 34 seconds… (attempt 7/10)
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 32 seconds… (attempt 8/10)
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 33 seconds… (attempt 9/10)
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 34 seconds… (attempt 10/10)
Same here... and what's more worrying is that there is everything green on status.claude.ai ... while it shouldn't...
Same here. Just started like 5-10 mins ago?
HELP, HOW DO I CODE NOW ? 😆
Confirmed that I only get
More H200s immediately!
Yeah it's been hit or miss although I seem to be getting some queries through right now after a few retries.
Seems like there's is* exponential backoff?
!Image
@ashwin-ant , I suggest that you may need to treat request based on how many tokens it have or expected server load needed
Example:
less than 2K tokens pass with no issue,
2k-4k backoff once
4k-6K backoff twice
6k-8K backoff three times
etc.. exponentially till it hit the 10 retries.
I think this will let the cute nice tasks pass and will not push all the users to the edge at once.
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 1 seconds… (attempt 1/10)
⎿ API Error (529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}) · Retrying in 1 seconds… (attempt 2/10)
Same problem here.
same here :(
same for me
same
same here
Same RN
same here
Opus 4 API is down now. Use Sonnet 4 instead.
https://status.anthropic.com/
having this issue at the moment
me too. maybe there are some issues on the server side. However, Sonnet4 is working now on Claude Code. I will just take a coffee break in the meantime until Opus4 comes back.
haa iyi gider aaaa :)
same here.
Having issue right now as well.
Having the same issue at the moment...
Just wait while monitoring the fixing progress through https://status.anthropic.com/
Take some break you hard workers :)
By using sonnet instead of opus, the error no longer occurs. I'll give up using opus until I get it back.
I confirm, using Sonnet works, /model > select sonnet
/model > Set model to sonnet (claude-sonnet-4-20250514)
This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.