What to know first
As of November 3, 2026: fast mode makes Opus up to 2.5x faster with the same model. Opus 5.5 fast mode costs $8 input and $40 output per million tokens, twice the standard $4/$20; Opus 5 and Opus 4.8 fast mode cost $10/$50. Subscribers must turn on usage credits under Settings → Usage, and fast mode spend does not count against plan limits. Before November 30, anyone still running fast mode on Opus 4.8 should either switch to Opus 5.5 with /model (Claude Code v2.1.280 or later) or run /fast off to return to standard speed.
What exactly changed?
The Claude Code 2.1.316 changelog (October 31, 2026) says: “Deprecated fast mode for Opus 4.8 as of October 30, 2026; it will be removed on November 30, 2026. After removal, fast mode requests on Opus 4.8 return an error and do not fall back to standard Opus 4.8.” The fast mode docs give the same dates and recommend migrating to Opus 5.5.
This is not a price increase, and standard-speed Opus 4.8 is unaffected. It affects people running /fast on Opus 4.8. The pattern matches Opus 4.7, whose fast mode was deprecated on June 25 and removed on July 24, 2026.
| Item | What Anthropic says |
|---|---|
| Supported models | Opus 5.5 (available); Opus 5 and Opus 4.8 (research preview); not Sonnet or Haiku |
| Opus 4.8 fast mode | Deprecated 2026-10-30, removed 2026-11-30; requests then error with no fallback |
| Speed | Up to 2.5x faster, same model and quality |
| How subscribers pay | Usage credits only, not included in plan limits |
| Not available on | Amazon Bedrock, Google Cloud Agent Platform, Microsoft Foundry, Claude Platform on AWS |
Sources: Claude Code Docs: Speed up responses with fast modeClaude Code Docs: Changelog (2.1.316, October 31, 2026)
What should Opus 4.8 users do before November 30?
The docs offer two paths. To keep fast mode, use /model to switch to Opus 5.5, which requires Claude Code v2.1.280 or later; fast mode stays on after the switch. To stay on Opus 4.8, run /fast off to return to standard speed. The same applies to direct API use: the Claude Platform fast mode docs say Opus 4.8 requests with speed: "fast" will return an error after November 30.
If a new session still starts on Opus 4.8, check the model and fastMode values saved in your settings. Opus 4.7 doesn't support fast mode, so switching to it turns fast mode off.
| Your situation | Action | Note |
|---|---|---|
| You want to keep fast mode | /model to Opus 5.5 | Needs Claude Code v2.1.280+; Opus 5.5 is the fast mode default from v2.1.280 |
| You want to keep Opus 4.8 | /fast off | Back to standard speed and pricing |
| You call the API with speed: "fast" | Change the model to Opus 5.5 or drop the speed parameter | Opus 4.8 fast requests error after 11-30 |
| You never turned on /fast | Nothing | Not affected |
Sources: Claude Code Docs: Speed up responses with fast modeClaude Platform Docs: Fast mode
How much more does fast mode cost than standard speed?
Fast mode has its own per-token price, flat across the 1M context window. Compared with Claude Platform's standard rates, fast mode on Opus 5.5 and Opus 4.8 costs twice as much. For how to do the token math, see our AI token cost example.
AISub's estimate at list price (excluding caching and data-residency multipliers): one Opus 5.5 request with 100,000 input tokens and 5,000 output tokens costs about $0.80 + $0.20 = $1.00 in fast mode, versus about $0.50 at standard speed. Another cost that is easy to miss: the first time you turn on fast mode mid-conversation, the whole context is billed at the uncached fast input price. With 300,000 tokens of context on Opus 5.5, that one request costs about 0.3 × $8 = $2.40. Fast and standard speed keep separate caches, so switching back and forth pays this again.
| Model | Standard input / output | Fast mode input / output | Multiple |
|---|---|---|---|
| Opus 5.5 | $4 / $20 | $8 / $40 | 2x |
| Opus 5 | $5 / $25 | $10 / $50 | 2x |
| Opus 4.8 (fast mode removed 11-30) | $5 / $25 | $10 / $50 | 2x |
Sources: Claude Code Docs: Speed up responses with fast modeClaude Platform Docs: Pricing
How do subscribers pay, and does it use Pro, Max or Ultra limits?
It doesn't use your plan limits, but it does cost real money. The docs say: “Fast mode usage draws directly from usage credits, even if you have remaining usage on your plan.” Until usage credits are on, /fast reports “Fast mode requires usage credits”. For how weekly limits work on each tier, see Claude Max weekly limits explained; for subscription prices, see the Claude price page.
Usage credits are billed at standard API rates and are separate from the monthly API credits included with Max and Ultra. Anthropic also sells prepaid usage bundles ($50 of usage for $45, $250 for $200, $1000 for $700), but the help center does not say whether bundle balance can pay for fast mode, so this article does not estimate discounted rates.
| Plan | Turn on usage credits | Where to see spend |
|---|---|---|
| Pro, Max, Ultra | Usage credits section of Settings → Usage on claude.ai, or run /usage-credits in Claude Code; if you subscribed through the mobile app, turn it on in the web app | Settings → Usage shows this month's usage-credit spend, including fast mode, not broken out |
| Team, Enterprise | A member with billing access turns it on at Organization settings → Usage; an Owner must also enable fast mode at Organization settings → Claude Code | Run /usage for your own spend |
| Claude Console (API key) | No usage credits; billed per token with other API usage | Console Usage/Cost pages, grouped by Speed |
Sources: Claude Code Docs: Speed up responses with fast modeAnthropic: Manage usage credits for paid Claude plansAnthropic: Buy usage bundles
When is it worth it, and how do you avoid surprise charges?
- Good fit: live debugging and rapid iteration, where waiting costs more than tokens. Poor fit: long autonomous tasks, batch jobs and CI.
- Run /status first. A login method such as “Claude Max account” means subscription plus usage credits; an API key row means Console billing.
- Set a monthly limit for usage credits under Settings → Usage, and think twice before turning on auto-reload.
- Teams can set "fastModePerSessionOptIn": true so every session starts with fast mode off, or disable it entirely with CLAUDE_CODE_DISABLE_FAST_MODE=1.
- When fast mode hits its own rate limit, it falls back to standard speed automatically. If usage credits run out, Claude Code retries at standard speed, and interactive sessions turn fast mode off.
- To compare Claude Code's subscription limits with other coding tools, see Codex vs Claude Code vs Antigravity.
Sources: Claude Code Docs: Speed up responses with fast modeAnthropic: Manage usage credits for paid Claude plans
Frequently asked questions
Is Claude Code fast mode included in a Pro or Max subscription?
No. On Pro, Max, Ultra, Team and Enterprise, fast mode is paid only through usage credits and is not included in plan limits. It draws from usage credits even if you have plan usage left.
Sources: Claude Code Docs: Speed up responses with fast mode
Can I still use Opus 4.8 after November 30?
Standard-speed Opus 4.8 is not affected by this change. What is removed is fast mode on Opus 4.8: after November 30, requests on Opus 4.8 with fast mode on are rejected and do not fall back to standard speed.
Sources: Claude Code Docs: Speed up responses with fast modeClaude Code Docs: Changelog (2.1.316, October 31, 2026)
How much does Opus 5.5 fast mode cost?
$8 per million input tokens and $40 per million output tokens, flat across the 1M context window, twice Opus 5.5's standard $4/$20. Fast mode stacks with caching and data-residency multipliers and is not available with Batch.
Sources: Claude Code Docs: Speed up responses with fast modeClaude Platform Docs: Pricing
Why is turning on fast mode mid-conversation so expensive?
The first time you turn it on in a conversation, the entire existing context is billed at the uncached fast input price. Fast and standard speed don't share a cache, so every later switch can pay again. That's why the docs recommend turning it on at the start of a session.
Sources: Claude Code Docs: Speed up responses with fast mode
Do usage credits expire?
The help center says usage credits generally don't expire; in Japan, credits purchased from September 10, 2026 expire six months after purchase. Credits work in Claude and Claude Code and are billed at standard API rates.
Sources: Anthropic: Manage usage credits for paid Claude plans
Official sources and review notes
Sources checked on November 3, 2026. Rates and eligibility rules can change; confirm the latest official guidance and your own agreement.
- Claude Code Docs: Speed up responses with fast mode
- Claude Code Docs: Changelog (2.1.316, October 31, 2026)
- Claude Platform Docs: Fast mode
- Claude Platform Docs: Pricing
- Anthropic: Manage usage credits for paid Claude plans
- Anthropic: Buy usage bundles
This guide draws on the public sources listed below. Examples illustrate the calculation methods. AISub is an independent price index and is not affiliated with the product providers discussed.