Errors and limits
Error format
Errors from the OpenAI-compatible endpoints use the OpenAI shape. The Anthropic endpoint returns Anthropic-shaped errors.
{
"error": {
"message": "The model 'modeldock/x' does not exist",
"type": "invalid_request_error",
"code": "model_not_found"
}
}Error codes
| HTTP | Code | Meaning |
|---|---|---|
| 401 | invalid_api_key | The key is missing, malformed, unknown or revoked. |
| 404 | model_not_found | No active model with that id on your account. |
| 403 | trial_ended | The Pre-flight for that model has ended. Launch again from the dashboard. |
| 400 | trial_prompt_too_long | A Pre-flight prompt is over 8,000 tokens. |
| 400 | max_tokens_exceeded | max_tokens is larger than the model's context length. |
| 429 | rate_limit_exceeded | Over the per-account request limit. Retry after the Retry-After seconds. |
| 504 | upstream_timeout | The model server timed out. Retry. |
| 503 | upstream_overloaded | The model server is busy. Retry after the Retry-After seconds. |
| 502 | upstream_error | The model server returned an error. Retry. |
Rate limit
Each account may make 10 requests per minute, shared across all its API keys. Responses carry x-ratelimit-limit-requests, x-ratelimit-remaining-requests and x-ratelimit-reset-requests headers, and a 429 includes Retry-After.
Model limits
- Up to 10 billion parameters.
- The Hugging Face repo must be public (not gated).
- Models that need custom remote code are not supported.
Pre-flight limits
A free Pre-flight ends at whichever of these comes first:
| Limit | Value |
|---|---|
| Live time | 15 minutes, counted from when the model is ready |
| Requests | 25 |
| Reply tokens in total | 30,000 |
| Reply tokens per request | 2,000 |
| Prompt size | 8,000 tokens |
| Completions per request | 1 (n=1) |
Each account gets 2 free Pre-flights. Further Pre-flights cost a flat $0.30 each, paid from your wallet. Launch attempts per day are also capped, failed ones included, because a failed launch still costs GPU start-up; the API answers 429 when you reach it.
More from litepod: the free LLM API with free open-source models. ModelDock is a litepod product.