Claude Code fast mode: what /fast costs and when it pays
Claude Code's /fast runs Opus up to 2.5x faster. On Opus 5.5 it costs $8/$40 per MTok, double standard, and needs usage credits. When it is worth it.
Fast mode is a switch in Claude Code that makes Claude Opus answer up to 2.5x faster in exchange for a higher price per token. Type /fast, confirm, and you are on. On Opus 5.5 it costs $8 per million input tokens and $40 per million output tokens, twice the standard $4 and $20 listed in our Opus 5.5 launch brief.
Anthropic's release notes say that on October 15, 2026 fast mode "is out of research preview for Claude Opus 5.5, and you no longer need to request access". That is the API side. Claude Code's own fast mode page still showed a research preview banner when we read it on October 20, so the two pages disagree.
This site is independent and not affiliated with Anthropic. We read the docs and did not run /fast ourselves, so nothing below is a hands-on result.
What fast mode changes, and what it does not
It does not change the model. The Claude Code docs put it plainly: "Fast mode is not a different model. It uses Claude Opus with a different API configuration that prioritizes speed over cost efficiency. You get identical quality and capabilities with faster responses."
It also only works on some models: "Fast mode is supported on Opus 5.5, Opus 5, and Opus 4.8. It is not available on Sonnet, Haiku, or other models."
The speed gain has a catch. Anthropic's API page says "Speed benefits are focused on output tokens per second (OTPS), not time to first token (TTFT)". Long answers arrive sooner. The wait before the first word is not what fast mode speeds up.
Turn it on
Check your plan
Fast mode needs usage credits on Pro, Max, Team and Enterprise. Team and Enterprise also need an Owner to enable it first.
Run /fast
Press Space to toggle it on, then Enter to confirm. Or set "fastMode": true in your user settings file.
Look for the confirmation
You should see "Fast mode ON" and a small ↯ icon next to the prompt.
If your current model is Sonnet, turning fast mode on moves you to Opus. The docs say "If your current model doesn't support fast mode, Claude Code switches to Opus". Turn it off with /fast again and you stay on Opus until you pick another model with /model.
Opus 5.5 is the fast mode default in Claude Code v2.1.280 and later, per the same page.
What it costs and who pays
Here are the Opus 5.5 rates per MTok, standard against fast:
| Input | Output | |
|---|---|---|
| Standard | $4 | $20 |
| Fast mode | $8 | $40 |
Fast mode on Opus 5 and Opus 4.8 is pricier still, at $10 and $50. The docs add that pricing is flat across the whole 1M token context window.
Here is the arithmetic on a made-up session of 50,000 input tokens and 15,000 output tokens, ignoring caching. Standard: $0.20 for input plus $0.30 for output, so $0.50. Fast: $0.40 plus $0.60, so $1.00. We computed this from the list prices. We did not measure a real session.
On a subscription, fast mode is not covered by your plan. The docs say "For Claude Code users on subscription plans (Pro/Max/Team/Enterprise), fast mode is available via usage credits only and not included in the subscription rate limits." If credits are off, /fast answers "Fast mode requires usage credits". For how the plan limits themselves work, see our Claude Code rate limits guide.
Our prompt caching guide explains why a mid-chat switch is expensive.
The Claude moves worth stealing, by email
Workflows tested on real jobs, sent only when they are worth it. Free.
When it pays, and when to skip it
Anthropic suggests fast mode for "Rapid iteration on code changes", "Live debugging sessions" and "Time-sensitive work with tight deadlines". It says standard mode is better for "Long autonomous tasks where speed matters less".
A simple rule: use it when you are watching the screen and waiting. Skip it when you start a job and walk away.
Fast mode is not the same as lowering the effort level. The docs describe them differently:
| Fast mode | Same model quality, lower latency, higher cost |
|---|---|
| Lower effort level | Less thinking time, faster responses, potentially lower quality on complex tasks |
You can use both together. Our guide to capping effort with maxEffortLevel covers the second control. For the cheaper model route, see Opus vs Sonnet.
Where it gets in your way
- Separate rate limits. The docs say "Fast mode has separate rate limits from standard Opus." When you hit them, Claude Code falls back to standard speed and the icon turns gray. Fast mode comes back by itself after the cooldown.
- Credits run out. Mid-session, Claude Code shows "Fast mode disabled · usage credits exhausted" and keeps working at standard speed and pricing.
- Admins can block it. Managed settings or a missing Owner opt-in produce "Fast mode has been disabled by your organization".
- Docs lag. The research preview label and the "up to 2.5x" figure come from Anthropic, not from us. Our 2.1.303-2.1.304 brief notes the changelog dropped the label from the
/fastdialog title. Check the page before you budget on it.
If you want a number for your own work, run one task each way and compare the usage figure on your plan. We have not done that.
How do I turn off fast mode in Claude Code?
Run /fast again, press Space to toggle it off, then Enter. You stay on Opus afterwards, and /model switches to another model. Setting the environment variable CLAUDE_CODE_DISABLE_FAST_MODE=1 disables it entirely.
Is Claude Code fast mode a different model?
No. Anthropic says "Fast mode is not a different model." It runs Claude Opus with a different API configuration, so quality and capabilities are the same. You get faster output at a higher price per token.
Does fast mode work on Sonnet?
No. The docs say fast mode "is not available on Sonnet, Haiku, or other models." It works on Opus 5.5, Opus 5 and Opus 4.8. Turning it on from Sonnet switches you to Opus first.
Is fast mode included in my Pro or Max plan?
No. On subscription plans it is "available via usage credits only and not included in the subscription rate limits." Turn on usage credits first. See our usage limits guide for what the plan itself includes.
Sources (checked 2026-10-20)
- Speed up responses with fast mode:
/fast, models, pricing, requirements, rate limits. - Fast mode on the API: up to 2.5x output tokens per second, Opus 5.5 availability.
- Claude Platform release notes: October 15, 2026 entry.
- Pricing: standard and fast mode rates per MTok.