pingoru pingoru SambaNova had 22 outages in the last 2 years totaling 83h 39m of downtime, averaging 0.9 incidents per month.
There were 22 SambaNova outages since October 29, 2024 totaling 83h 39m of downtime. Each is summarised below. Incident details, duration, and resolution information.
SambaNova Cloud login authentication is not functional. Call with your existing API keys are fine.
The issue has been identified and a fix is being implemented.
This incident has been resolved.
We are currently having issues with E5-Mistral-7B-Instruct down . All other models are responding fine
E5-Mistral-7B-Instruct IS back in service
SambaNova Cloud Down we are investigating the issue
The issue has been identified and is process of recovering
The issue has been resolved.
The team is actively investigating the issue .
The issue is now resolved. We will continue to monitor for re-occurances.
We are currently investigating this issue.
The Issue has been resolved. All three models are fully operational.
Model gpt-oss-120b may return 504 for certain sequence lengths
The incident with model gpt-oss-120b has been resolved.
We are currently experiencing issues with access to Qwen3-32B.
Access to Qwen3-32B has been restored.
You may experience errors when accessing Llama-4-Maverick-17B-128E-Instruct via API . We are actively investigating.
Resolved: Llama-4-Maverick-17B-128E-Instruct partial outage is now resolved. Contact [email protected] if further details are required.
We are presently experiencing high demand impacting performance on Meta-Llama-3.1-8B-Instruct.
The issue has been resolved and Meta-Llama-3.1-8B-Instruct response should be back to normal.
Between approximately 08:37 - 09:18 PDT we suffered an access interruption to Deepseek, Llama 405B, QwQ,Mistral, and Whisper. Service was restored and the root cause is known and will be corrected.
We are currently investigating and issue which may cause a disruption in service.
We located the issue and all services are up and available.
Authentication into our SambaNova Cloud portal as well as Authentication into our SambaNova Developer Community are presently impacted. This will not cause an impact if your are already authenticated into those sites. It will not cause impact to any API calls. If you require immediate assistance email [email protected].
Authentication to SambaNova Cloud portal and SambaNova Developer Community Restored
We have put 16k into maintenance mode smaller context lengths are still avialable.
DeepSeek R1 16k is available again and we are actively monitoring.
This incident has been resolved.
We are presently debugging an issue with our Playground and API
We have Identified the issue
A fix has been implemented and we are monitoring the results.
This incident has been resolved.
We are currently triaging an issue with confirmed impact in India and Japan
The issue is resolved
The issue has been identified and a fix is being implemented.
Issue is resolved.
At present certain logins to the Japan based authentication services are receiving a nonce error.
Moving to partial outage as it is just effecting Auth
The Issue is known and a workaround is in place until a permanent fix goes in via the official release process.
This incident has been resolved.
We are currently investigating the issues with Meta-Llama-3.1-405B-Instruct
We have identified the issue and put in a preventative measure . We will continue to monitor.
This incident has been resolved.
The issue has been identified and a fix is being implemented.
A fix has been implemented and we are monitoring the results.
We are continuing to monitor for any further issues.
This incident has been resolved.
Llama 3.1 8B is having intermittent timeouts and we are investigating.
Llama 3.1 8B is having intermittent timeouts has been resolved.
We are currently investigating this issue.
The issue has been identified
The issue appears to have abated and we are monitoring
We are continuing to monitor for any further issues.
We are continuing to monitor for any further issues.
Consistant service to all models is restored
This incident has been resolved.
Our team is looking into an issue where a few models are unresponsive in the Playground.
The issue has been identified and our team is working on a fix.
The issue has been resolved. Our team will monitor for further concerns.