Is Fireworks AI down?
No, Fireworks AI is not down.
Fireworks AI’s status page reports no incidents, api.fireworks.ai responded from 6 regions, and nobody reported problems in the last hour.
- Official statusNo incidentsstatus.fireworks.ai
- Reachable6 of 6 regionsFireworks AI API · api.fireworks.ai
- Reports in the last hour0Normal
Fireworks AI’s status page
From status.fireworks.ai. Read at
- MiniMax M3.1 FlashOperational
- DeepSeek V4.1 Flash 1017Operational
- DeepSeek V4.1 Flash 1017 USOperational
- Muse Spark Open USOperational
- GLM 5.3 FlashOperational
- GLM 5.3 Flash USOperational
- GLM 5.3Operational
- GLM 5.3 FastOperational
Show all 22 components
- GLM 5.3 USOperational
- Ember-1Operational
- Kimi K3 USOperational
- Kimi K3 FastOperational
- Kimi K3Operational
- Qwen 3.8 MaxOperational
- Deepseek V4.1 FlashOperational
- Deepseek V4.1 Flash USOperational
- Nemotron 3 Ultra USOperational
- Nemotron 3.5 Lightning USOperational
- Minimax M3 USOperational
- MiniMax M3Operational
- InklingOperational
- GPT-OSS 120BOperational
Not working for me
Pick what’s broken. Reports from everyone show up on the chart below within 5 minutes.
Last 24 hours
No problems in the last 24 hours, median response 95 ms, and 0 reports.
- Operational
- Degraded
- Partial outage
- Major outage
- No data
- Reachability check (api.fireworks.ai)
By region
| Region | Fireworks AI API api.fireworks.ai |
|---|---|
| N. Virginia us-east-1 | 401 · 26 ms |
| Oregon us-west-2 | 401 · 13 ms |
| Frankfurt eu-central-1 | 401 · 26 ms |
| Singapore ap-southeast-1 | 401 · 38 ms |
| Tokyo ap-northeast-1 | 401 · 37 ms |
| São Paulo sa-east-1 | 401 · 382 ms |
Two regions check Fireworks AI every minute, and any failure is rechecked from all of them. Each row shows the latest result. How we check
Last 90 days
We have checked Fireworks AI ourselves since Oct 28, 2026. Earlier days come from its status page.
42 days with problems in the last 90 days.
Recent incidents
Incidents: 50 in the last 30 days and 50 in the last 90, lasting a median of 17 min.
- Service Degradation for GLM 5.3 FlashDegraded23 min
- Service Degradation for Kimi K3Degraded30 min
- Service Degradation for Kimi K3Degraded7 min
- Service Degradation for GLM 5.3 FlashDegraded23 min
- Service Degradation for Kimi K3Degraded2 h 34 min
- Service Degradation for MiniMax M3.1 FlashDegradedunder 1 min
- Service Degradation for Kimi K3Degraded7 min
- Service Degradation for Kimi K3Degraded14 min
- Service Degradation for GLM 5.3Degraded4 min
- Service Degradation for GLM 5.3Degraded3 h 5 min
If Fireworks AI isn’t working for you
Fireworks AI is an inference platform that serves open-source and fine-tuned models on serverless and dedicated GPUs.
Common Fireworks AI errors
| Error | What it means | Usually | What to do |
|---|---|---|---|
429 Too Many Requests | On serverless, you exceeded your request or token limit; on a dedicated deployment, its GPUs are saturated. | Heavy load | Retry with backoff and smooth bursts; for dedicated deployments, add replicas. Source |
503 Service Unavailable | The service is down for maintenance or experiencing issues. | An outage | Retry after some time and check status.fireworks.ai. Source |
502 Bad Gateway | The server got an invalid response from an upstream server. | An outage | Wait and retry; if it persists it may indicate an outage. Source |
402 Payment Required | The account is not on a paid plan or exceeded its usage limits. | Your side | Check billing status and payment method. Source |
412 Precondition Failed | The account is suspended or has a status problem, or a LoRA model failed to load. | Your side | Check account status and billing, or re-upload the LoRA model. Source |
401 Unauthorized | The API key is invalid or lacks permissions. | Your side | Verify the API key and its permissions. Source |
Is it just you?
- On a dedicated deployment, a 429 means your GPUs are full, not that Fireworks is down; check the deployment's metrics.
- Try the same request against a serverless model to separate your deployment from the platform.
- Check billing and account status for 402 and 412 errors.
- Check status.fireworks.ai and the reports on this page.
Alternatives to Fireworks AI
Questions
On serverless it means you exceeded your request or token limit. On dedicated or on-demand deployments, Fireworks has no account-level rate limits, so a 429 means the deployment is at capacity.
Check the live status at the top of this page, which combines Fireworks' official status with our own checks. Fireworks posts incidents at status.fireworks.ai.
Sources: docs.fireworks.ai, status.fireworks.ai
Get alerted when Fireworks AI breaks your app.
If your product or your team depends on Fireworks AI, its outage is your outage, and your users will blame you, not Fireworks AI. Point a Site Qwality monitor at the Fireworks AI endpoint you actually rely on, put your own site next to it, and we alert you on Slack, email, SMS or any webhook the moment either one stops answering.