Is AWS Bedrock Down?
Live AWS Bedrock status. Auto-refreshes every 2 minutes.
AWS Bedrock is Degraded
AWS Bedrock is experiencing degraded performance. Some features may be slower or intermittently unavailable.
Official status page: health.aws.amazon.com/health/status
Component Status
Recent AWS incidents
What to do when AWS Bedrock is down
Call the model owner directly
Because Bedrock is a managed gateway in front of other companies' models, the fastest fallback is often the model provider's own API. If you were calling Claude on Bedrock, hit the Claude API directly. If you were calling Mistral, use Mistral's own endpoint. The model behaves the same, you just bypass the AWS layer that is having the event. Keep API keys for one or two model owners ready so you can switch without a code rewrite.
Route through a sibling inference gateway
Other multi-model gateways host overlapping catalogs (Llama, Mistral, Qwen, DeepSeek and more) on separate infrastructure, so a Bedrock event does not touch them. OpenRouter can fail over across providers automatically, while Together AI, Fireworks, and Groq run the open-weight models on their own stacks. Compare current uptime across gateways before you re-point traffic.
Get notified when status changes
Skip the manual refreshing. TensorFeed can alert you the moment Bedrock or any tracked AI provider flips between operational, degraded, and down, so you find out before your users do. You can also watch every major provider on one live board to decide where to send traffic during an incident.
Frequently Asked Questions
Is AWS Bedrock down right now?
The live indicator at the top of this page is the answer. Green means no active AWS Health events are affecting Bedrock in the regions we monitor. Amber means degraded performance in at least one region, and red means a significant outage. Because Bedrock is regional, always read the affected-region list below the headline, since a problem in us-east-1 does not mean us-west-2 or eu-central-1 is impacted.
How do I check AWS Bedrock status?
This page pulls the AWS Health feed every two minutes and filters it down to events affecting Bedrock specifically, which is faster than scanning the full AWS dashboard by hand. The official source is the AWS Health Dashboard at health.aws.amazon.com/health/status, where you can filter by the Bedrock service and by region. For account-specific impact, the personalized AWS Health Dashboard inside your console shows events tied to the exact resources you are running.
What should I do when AWS Bedrock is down?
First check whether the event is regional. The standard Bedrock failover pattern is to retry in another region where the same model is enabled, since model availability and outages are per region. If the whole service is impaired, fall back to calling the model owner directly (for example the Claude API or Mistral's own API) or route through a sibling gateway. Build exponential backoff and a cross-region or cross-provider fallback into your client so a single region's event does not take your app down.
How often does AWS Bedrock go down, and how reliable is it?
Bedrock is generally reliable and most weeks see no customer-facing outage, but it inherits the blast radius of core AWS infrastructure. Large regional AWS events (networking, IAM, or DynamoDB issues in a region like us-east-1) can cascade into Bedrock invocation errors and throttling even when the model layer itself is fine. Throttling and capacity limits during peak demand are a more common day-to-day issue than full outages, especially for newer or high-demand models, which is why provisioned throughput exists.
Which models and services does AWS Bedrock actually run?
Bedrock is a managed gateway, not a single model. It serves Anthropic Claude, Meta Llama, Mistral and Mixtral, Amazon's own Titan and Nova families, Cohere, AI21, and Stability image models, plus features like Knowledge Bases, Agents, Guardrails, and fine-tuning. A common confusion is mixing it up with Amazon SageMaker (which is for training and hosting your own models) or assuming a model is available everywhere. Model and feature availability varies by region, so an outage or limit may apply to one model in one region while the rest of Bedrock runs normally.
Where can I see AWS Bedrock incident history?
AWS keeps a public event log on the AWS Health Dashboard at health.aws.amazon.com/health/status, where you can browse recent and historical events filtered by service and region. For incidents that hit your own workloads, the personalized AWS Health Dashboard in the console retains your account's event history. Major regional AWS outages are also written up in detailed post-event summaries that AWS publishes after the fact.