Skip to content

Free LLM API errors · OpenRouter · checked 2026-11-13

OpenRouter empty response with HTTP 200: why

HTTP 200 · HTTP 200, but the reply is empty or the stream ends with finish_reason "error"

OpenRouter sends 200 as soon as a provider accepts the request. Anything that goes wrong after that is reported inside the response, not in the status code.

Our own router has received this error from OpenRouter in production.

Why it happens

  • A provider error during generation: streaming responses get an SSE chunk with a top-level error object and finish_reason "error", then the stream ends.
  • Non-streaming requests can get a 200 whose body has only an error object and no choices.
  • Reasoning models can spend the whole output on hidden reasoning and return no visible text.

How to fix it

  1. Check for an error field on every chunk and on every 200 body, not only on non-200 statuses.
  2. Treat finish_reason "error" and an empty answer as a failure and retry, or move to another model.
  3. OpenRouter stops failing over once part of the answer has reached you, so a mid-stream failure is yours to handle.
  4. For reasoning models, ask for less reasoning (OpenRouter's reasoning.effort) or allow more output tokens.

Detect an error inside a 200 stream (JavaScript)

for await (const chunk of stream) {
  if (chunk.error) throw new Error(chunk.error.message);
  const choice = chunk.choices?.[0];
  if (choice?.finish_reason === "error") throw new Error("provider failed mid-stream");
  text += choice?.delta?.content ?? "";
}
if (!text.trim()) throw new Error("empty response");

Questions

Why does OpenRouter return 200 for a failed request?
The 200 is sent when the provider accepts the request. Failures after that arrive in the body or as an SSE event.
Why is the reply from a reasoning model empty?
It can use all its output on reasoning that is not shown. Lower the reasoning effort or raise the output limit.

Sources

Read on 2026-11-13. Providers change limits and messages often; the linked pages are what applies.

Limits and fine print of OpenRouter's free tier: OpenRouter free API. Tired of handling these by hand? ZeroLimitAI is one OpenAI-compatible endpoint that moves to the next free model when one of these comes back, and can run on your own Groq or OpenRouter key.

Other OpenRouter errors

Stop paying per token.

One endpoint. $0 inference.

Free 7-day API key in a minute · no card. Then $99 once on your own key, or $99 a year.