Flash Answers

Flash Answers is a fast-response mode that delivers answers at up to 10x the usual speed. When you need a quick answer, a first draft, or rapid iteration and don't need step-by-step reasoning, Flash Answers gets you there faster. It generates over 1,000 tokens (roughly 750 words) per second, so responses feel almost instant.

How it works

How it works

Flash Answers prioritizes speed over reasoning depth. Instead of working through a chain-of-thought process, it gives you a direct answer as quickly as possible.

Le Chat uses Flash Answers automatically in certain features. For example, Deep Research relies on it to generate full reports at speed once the research phase is complete.

How to activate

How to activate

  1. Click the Flash button at the bottom of the chat window.
  2. Type your message and send it as usual.

Flash Answers stays active for all subsequent messages in the conversation until you turn it off. Click the Flash button again to switch back to the standard model.

When to use Flash Answers

When to use Flash Answers

Flash Answers works best when speed matters more than depth:

  • Quick factual questions: look up a definition, check a date, or get a concise explanation during a meeting.
  • Short-form content generation: draft a quick email, write a social media post, or outline talking points.
  • Iterative brainstorming: generate and refine ideas rapidly without waiting between responses.
  • First drafts: get a rough version on screen fast, then switch to the standard model or Canvas to polish it.
Limitations

Limitations

Flash Answers trades reasoning depth for speed. It isn't the best choice for tasks that require multi-step logic, nuanced analysis, or careful evaluation of trade-offs. For those, use Think mode, which applies chain-of-thought reasoning to work through complex problems step by step.

tip

If your question has a straightforward answer, use Flash Answers. If you'd want to see the model's reasoning before trusting the answer, use Think mode instead.

  • Think mode: step-by-step reasoning for complex, multi-factor problems.
  • Deep Research: automated multi-source web research with structured reports (uses Flash Answers for fast generation).
  • Chat: standard conversational mode with full model capabilities.