Is Cohere Down?
Live Cohere status. Auto-refreshes every 2 minutes.
Cohere is Operational
Cohere is up and running normally. All systems are operational.
Official status page: status.cohere.com
Component Status
Recent Cohere incidents
What to do when Cohere is down
Swap Command for another generation API
If you call Cohere's Command models for chat or text generation, a comparable frontier API can cover you during an outage. Claude and GPT both handle instruction following, summarization, and tool use, and most SDKs let you switch the base model behind a single config flag. Check that the alternative is itself healthy before you cut over.
Reach Cohere models through a gateway or pick an alternate for Embed and Rerank
Cohere's Command and Embed models are also served through multi provider gateways, so routing through one can sidestep a direct API problem or give you instant fallback to a different model. For embeddings and reranking specifically, open weight and hosted alternatives on inference platforms can stand in until Cohere recovers.
Get notified when status changes
Rather than refreshing this page, set up alerts so you hear the moment Cohere's status flips to degraded or down, and again when it recovers. TensorFeed can also notify you across the other AI providers you depend on, so a single outage does not catch your whole stack by surprise.
Frequently Asked Questions
Is Cohere down right now?
The live indicator at the top of this page reflects Cohere's current state, pulled from monitoring and the official Cohere status page. A green dot means the API is operational, amber means degraded performance (slower responses or intermittent errors), and red means a confirmed outage. If you see green here but your own requests are failing, the problem is more likely your API key, your rate limit, or a specific endpoint than a platform wide outage.
How do I check if Cohere is down?
Start with the live indicator on this page, then confirm against Cohere's official status page at status.cohere.com, which breaks status out by component such as the Chat (Command) API, Embed, and Rerank. From your own side, check the HTTP status code your client returns: a 401 or 403 points to authentication, a 429 means you hit a rate limit or quota, and 500 or 503 codes suggest a server side problem on Cohere's end.
What do I do when Cohere is down?
First confirm it is Cohere and not your key or quota by checking the returned status code and Cohere's status page. If it is a real outage, add retry logic with exponential backoff for transient 5xx errors, since most incidents resolve in minutes. For longer outages, route to a fallback provider: if you are using Command for generation you can swap to Claude or GPT, and for Embed or Rerank you can fall back to another embeddings or reranking provider while you wait.
How often does Cohere go down, and how reliable is it?
Cohere is an enterprise focused provider and is generally stable, with most disruptions being short lived degraded performance rather than full outages. Incidents do happen, typically tied to elevated latency, a specific model or endpoint, or upstream cloud issues. Cohere publishes incident history and component uptime on its status page, which is the most accurate source for actual reliability over time rather than any single number quoted elsewhere.
Which Cohere services and models could be affected during an outage?
Cohere's API spans several distinct products that can fail independently: the Command family for chat and generation, Embed for turning text into vectors, and Rerank for reordering search results by relevance. An incident may hit only one of these, so a Rerank slowdown does not necessarily mean Embed or Command are affected. Cohere is also offered through cloud marketplaces such as Amazon Bedrock, Microsoft Azure, and Oracle Cloud, and on those platforms availability can depend on the cloud provider's own status as well as Cohere's.
Where can I see Cohere's incident history?
Cohere's official status page at status.cohere.com lists past incidents, maintenance windows, and per component uptime, which is the canonical record. For broader context across the AI ecosystem, TensorFeed tracks Cohere alongside other providers so you can see whether an issue is isolated to Cohere or part of a wider pattern affecting multiple APIs at once.