Is the Kimi API down?
The API rows from Moonshot AI's status page are below. During an overload the Kimi API returns HTTP 429 "engine_overloaded_error" or HTTP 503 "server_unavailable". Retry with backoff and honor Retry-After, switch model, or run Kimi K3 on another host: the weights are open.
Live Kimi API status
Checked
All systems operational
- Open APIOperational
- Open Platform PortalOperational
- Last incident: 开放平台 API 服务异常resolved Sep 2
Source: status.moonshot.cn, API rows and API incidents only; Moonshot AI writes incident titles in Chinese. Read on our server about every 2 minutes.
Which error is Moonshot, and which is you
| What you see | What it means |
|---|---|
HTTP 429, engine_overloaded_error | "The engine is currently overloaded, please try again later": Moonshot's capacity. Topping up does not help; follow Retry-After |
HTTP 503, server_unavailable | Temporarily unavailable, usually scaling or maintenance at Moonshot |
HTTP 500, server_error | An internal error. Retry, then check the rows above |
| HTTP 504 Gateway Time-out | No answer within 900 seconds; the body is an HTML page |
HTTP 429, rate_limit_reached_error | Your account's concurrency, RPM, TPM or TPD limit |
HTTP 429, exceeded_current_quota_error | Your balance is too low or the account is disabled |
HTTP 401, invalid_authentication_error | Your key. Keys do not work across Moonshot's regional platforms, and Kimi Code keys are not Open Platform keys |
Five minutes into a Kimi outage
- Check the Open API row above. The consumer Kimi app has its own rows, and its incidents do not always touch the API.
- Retry with exponential backoff and honor Retry-After. Moonshot does not charge requests interrupted by a 429.
- If it lasts, switch to another model, or reach Kimi K3 through another host of the open weights.
How to switch in your agent
| Agent | Steps |
|---|---|
| Kimi Code CLI | Add a second provider under [providers.<name>] in ~/.kimi-code/config.toml with type = "openai", its base_url and api_key_env, then switch with kimi -m <alias> |
| Cline, ZooCode, Kilo Code | Open the extension's settings, change API Provider or Model, and save. The next request uses it |
| OpenCode | Type /models and pick a model from another provider, or start with opencode -m provider/model |
| Aider | Type /model followed by another model name; the chat history stays |
| Your own code | Catch 5xx and 529, retry with exponential backoff, then send the request to a fallback model |
What Standard Compute does when Moonshot fails
- Every request goes through OpenRouter. Depending on the key, Kimi K3 goes to Moonshot AI or Modal first; if that host fails, OpenRouter can send the request to another host of the same model.
- A key restricted to the US runs Kimi K3 only on approved US endpoints (DeepInfra and Fireworks), not on Moonshot's own API. Kimi K3 has no approved endpoint in Europe.
- Each failed call is retried twice on a 429, a 5xx or a network error, after 0.5 and then 1 second, or after the provider's Retry-After up to 8 seconds. If every attempt fails, your agent gets HTTP 503 with "The model is temporarily unavailable. Please retry shortly."
- A specific model ID always stays that model: failover changes the provider, never the model.
The catch
Other hosts run the same Kimi K3 weights, several at FP4 or FP8 precision, and new Kimi releases can reach them later than Moonshot's own API. If Moonshot is down at the model level, for example during a bad release, every host of that model may be affected.
Where Standard Compute fits
Standard Compute is built to make coding agents cost less: its router sends each step to the cheapest model that can do it, on providers we vet for price, uptime and speed, and a failed provider falls back automatically to another provider of the same model. An opt-in Compliance rule keeps a key, a team or the whole company on endpoints in Europe, the United States or both, and we don't train on your prompts or store them. One key at https://api.stdcmpt.com/v1 serves Claude, GPT, Gemini, DeepSeek, GLM and Kimi models, from $19 a month (USD). Kimi K3 is moonshotai/kimi-k3.
FAQ
Is Kimi down or is it just me?
Check the Open API row on status.moonshot.cn. If it is green, read the error type: invalid_authentication_error is your key, rate_limit_reached_error your account limit, exceeded_current_quota_error your balance, and engine_overloaded_error Moonshot's capacity.
What does engine_overloaded_error mean on the Kimi API?
Moonshot's servers are over capacity. Moonshot's docs say topping up or a higher tier does not fix it. Retry with backoff and follow Retry-After; requests interrupted by a 429 are not charged.
Can I use Kimi K3 when Moonshot's API is down?
Yes. Kimi K3's weights are published under the Kimi K3 License, and other providers host them. On Standard Compute a Kimi K3 request can fail over from Moonshot to another host of the same model.
Why are the incidents on Moonshot's status page in Chinese?
Moonshot AI writes them that way. The page also covers the consumer Kimi app; for the developer API, read the Open API and Open Platform Portal rows.