---
title: A reply stopped before it finished
type: troubleshooting
section: ai-studio
summary: A reply that ends early carries a note saying who ended it. The note tells you whether to send again, lower the reasoning level, or change model.
applies_to:
  clients: [web]
sources:
  - evolve-front-end/app/[lng]/studio/components/Chat/AssistantTerminalNotice.tsx
  - evolve-front-end/app/[lng]/studio/components/Chat/lower-effort-retry.ts
  - evolve-front-end/app/[lng]/studio/components/Stream/studio-stream-reducer.ts
  - evolve-front-end/app/[lng]/studio/components/Chat/RawSseSections.tsx
  - evolve-front-end/app/i18n/locales/en/powerflow.json
verified:
  at: 2026-09-04
  env: code
  by: content-lane-ai-studio
ttl_days: 45
related: [ai-studio/token-limits, ai-studio/choose-a-model, ai-studio/cost-and-usage]
---

A reply that ends early carries a note under it. For the author of the thread the note names what ended the reply; anyone else reading the thread sees only the generic wording. Read the note first: it decides what to do next.

## The AI provider ended this response early. The text above may be incomplete.

The model stopped on its own, most often because the answer ran past the length the provider allows in one response.

1. Send a follow-up asking it to continue from where it stopped.
2. Ask for a shorter answer, or break the question into parts.
3. Where the reply used a high reasoning level, select **Retry with lower reasoning** below the note. It appears when a lower level is available on that model.

## The AI provider ended the response before returning any text.

The model produced nothing at all. Send the message again. If it repeats, change model: see [Choose a model](/help/ai-studio/choose-a-model).

## AI Studio could not complete this response.

The failure was on the Evolve side, not the model's. Send the message again. If it repeats, note the reference shown under the message and contact Support with it.

## A tool could not complete this response.

A tool the model called did not return. Where you have added an MCP server to the thread, check it is reachable and that the tool you expect is allowed: see [Connect an MCP server](/help/ai-studio/connect-an-mcp-server). Then send the message again.

## You stopped this response.

You selected the stop button. The partial text is kept. Send the message again to start over.

## The response ended unexpectedly.

Either the cause was not recorded, or you are reading a thread you did not write; a reader who is not the author sees this in place of the detail. Open your own thread to see the full note, or send the message again.

## Was I charged for it?

The author of the thread sees a billing line under the note: whether the incomplete reply was charged, is still being confirmed, or needs review. See [What your requests cost](/help/ai-studio/cost-and-usage).

## Related

- [Token limits in a thread](/help/ai-studio/token-limits) — the separate limit that stops a thread accepting new messages.
- [Choose a model](/help/ai-studio/choose-a-model) — a message that fails on two different models needs to be shorter, or a different model.
- [What your requests cost](/help/ai-studio/cost-and-usage) — whether an incomplete reply was charged.
- If nothing above matches, copy the reference shown under the note and contact Support with it.