Claude Opus 4.8 Fast Mode: 2.5x Faster Output Tokens; Opus 5.5 Is Now the Default
Anthropic launched fast mode for Claude Opus 4.8 on May 28, 2026: the same model at 2.5x the output speed, three times cheaper than before. Opus 5.5 is now the fast mode default in Claude Code v2.1.280 and later, at $8/$40 per MTok; Opus 4.8 still supports it at $10/$50, billed to usage credits on subscription plans.
Status (October 9, 2026): Fast mode now defaults to Opus 5.5 in Claude Code v2.1.280 and later, at $8/$40 per MTok. Opus 4.8 still supports it at $10/$50, billed to usage credits on subscription plans.
Anthropic launched fast mode for Claude Opus 4.8 in research preview on May 28, 2026. The pitch was "the same Opus-level model intelligence with significantly faster speeds," quantified as "2.5x faster output token speeds" compared to standard Opus 4.8. In Claude Code you turn it on with /fast; on the Claude Platform API, access goes through an account manager or a waitlist.
This extends the Fast mode story we covered when Claude Code 2.1.142 moved Fast mode's default to Opus 4.7; 2.1.154 then made Opus 4.8 the default, and it stayed there until v2.1.219. For the model generation context, see the Opus 4.7 launch, and for how Claude Code stacks up against the alternative most readers weigh it against, our Cursor vs Claude Code comparison is the anchor.
Key Takeaways
- 2.5x faster output tokens. At launch, Anthropic's Fast mode page cited 2.5x faster output token speeds for Opus 4.8 versus standard Opus 4.8; the page now describes the same speedup for Opus 5.
- Same intelligence, more speed. Fast mode is positioned as "the same Opus-level model intelligence with significantly faster speeds," not a smaller or distilled model.
- Research preview, not GA. Anthropic describes Fast mode as a research preview with restricted availability.
- Claude Code: toggle with
/fast, billed to usage credits. On subscription plans fast mode draws only on usage credits, which must be turned on; Team and Enterprise need an Owner to enable it. Since v2.1.280 the default fast mode model is Opus 5.5. - API: account manager or waitlist. Claude Platform API access is a limited research preview; contact your account manager or join the waitlist form linked from the page.
- Cheaper than before, still a premium. Anthropic's announcement says it made fast mode "three times cheaper than before," and the Claude Code docs list it at $10/$50 per million input/output tokens on Opus 4.8 and $8/$40 on Opus 5.5, "at a higher cost per token" than standard Opus.
What Fast Mode Does for Opus 4.8
Fast mode is Anthropic's option to run Opus 4.8 with substantially faster output without dropping to a smaller model. At launch, the Fast mode page described it as delivering "the same Opus-level model intelligence with significantly faster speeds," and for Opus 4.8 it put a number on the speed: "2.5x faster output token speeds" compared to standard Opus 4.8. The headline benefit is latency on long generations -- the part of an agent loop where you wait on tokens streaming back.
That framing matters because speed gains often come from a weaker model. Anthropic's claim here is the opposite: the intelligence is the same Opus generation, and the lever is throughput. For interactive coding, faster output tokens translate directly into shorter waits between a prompt and a usable diff.
How to Enable It in Claude Code
In Claude Code you switch fast mode on with the /fast command. At launch Anthropic described it as "available now in research preview for all developers with extra usage enabled." The fast mode docs now call that prerequisite usage credits: on Pro, Max, Team and Enterprise plans "fast mode is available via usage credits only and not included in the subscription rate limits," and /fast reports "Fast mode requires usage credits" until they are turned on. Team and Enterprise organizations also need an Owner to enable it.
Opus 4.8 held the fast mode default from v2.1.154 through v2.1.218. The docs now say "Opus 5.5 is the fast mode default in Claude Code v2.1.280 and later. Before v2.1.280, fast mode defaulted to Opus 5 from v2.1.219, to Opus 4.8 on v2.1.154 through v2.1.218". Fast mode "is supported on Opus 5.5, Opus 5, and Opus 4.8," so to keep using it on Opus 4.8, switch models with /model. Fast mode on Opus 4.7 was removed on July 24, 2026.
| Model | Fast mode price per MTok (input/output) |
|---|---|
| Opus 5.5 (default from v2.1.280) | $8 / $40 |
| Opus 5 | $10 / $50 |
| Opus 4.8 | $10 / $50 |
How to Get It on the API
API access is more gated. Anthropic describes the Claude Platform path as a "limited research preview." Access "requires contacting your account manager," and teams without an account manager can "join the waitlist for fast mode by completing this form" linked from the Fast mode page. In other words, the Claude Code path is self-serve for accounts with usage credits turned on, while the raw API path is request-based. The Fast mode page now describes the API waitlist in terms of Opus 5.
Availability
The table below summarizes what Anthropic documents on the Fast mode page and in the Claude Code docs.
| Surface | Status | How to access |
|---|---|---|
| Claude Code | Research preview, available now | Usage credits turned on; toggle with /fast (Opus 5.5 by default) |
| Claude Platform API | Limited research preview | Contact account manager, or join the waitlist form |
When to Use It
Reach for fast mode when latency is the bottleneck: long code generations, agentic loops with many turns, or any workflow where you sit watching tokens stream. Because Anthropic positions it as the same intelligence at higher speed, there is no quality reason to avoid it for those cases. On a current Claude Code release it runs on Opus 5.5 by default, which is also the cheapest fast mode option at $8/$40 per million input/output tokens; stay on Opus 4.8 only if you have tuned prompts to it. The gating factors are practical -- usage credits turned on in Claude Code, or an account-manager relationship or waitlist slot for the API. At launch Anthropic made fast mode three times cheaper than before, but it still carries "a higher cost per token" than standard Opus, so factor that premium in before standardizing a team on it. For prompt changes when moving up, see prompting Claude Opus 5.5.
Corrections and Updates
- October 9, 2026: This page said Opus 4.8 fast mode was the current option, enabled with extra usage; now Opus 5.5 is the fast mode default from v2.1.280, at $8/$40 per MTok, and subscription plans pay through usage credits.
- October 9, 2026: This page said Anthropic's pricing "was cut three times over"; the announcement said fast mode was made "three times cheaper than before."
Sources
- Anthropic, "Fast mode" page: https://claude.com/fast-mode
- Claude announcement post (May 28, 2026): https://x.com/claudeai/status/2060042706844315866
- Claude Code docs, "Fast mode" (fetched October 9, 2026): https://code.claude.com/docs/en/fast-mode
- Claude Code changelog, 2.1.154 ("Fast mode on Opus 4.8 is now available at a fraction of its previous cost") and 2.1.280 (Opus 5.5 added): https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md
Read next
Related coverage- Claude Code
Claude Code 2.1.142: `claude agents` Gains Session Flags, Fast Mode Defaults to Opus 4.7, MCP Tool Timeout Honored
Claude Code 2.1.142 (May 14, 2026) added eight dispatch flags to `claude agents`, moved fast mode's default from Opus 4.6 to Opus 4.7, and made `MCP_TOOL_TIMEOUT` lift the 60-second cap on remote MCP servers. The flags and timeout fix stand; fast mode now defaults to Opus 5.5 (v2.1.280 and later) at $8/$40 per MTok.
- Claude
Claude Opus 5.5: Fable 5.1-Level Scores at $4 and $20, With Benchmarks, Effort Costs, and Migration Guide
Claude Opus 5.5, released September 22, 2026, beats Claude Fable 5.1 on every benchmark Anthropic published and costs $4 and $20 per million tokens, 60% less than Fable 5.1. It scores 66.4% on Terminal-Bench 4.0 against 57.9% for GPT-6 Astra. Thinking is always on, the default effort is medium, and four API changes can break Opus 5 code.
- Claude
Prompting Claude Opus 5.5: Delete 'Think Carefully', Name the Finish Line, Start at Medium
Prompt Claude Opus 5.5 by handing over the whole task with a clear finish line, deleting “think carefully” lines, and starting at medium effort. Thinking is always on, so effort is your depth control. Add a CLAUDE.md rule that names when to stop, and set effort explicitly instead of carrying Opus 5 settings over.
- Claude
Claude Opus 4.7 Launch: Stronger Coding, xhigh Effort, and a New Cyber Safeguards Tier
Anthropic launched Claude Opus 4.7 on April 16, 2026, with gains over Opus 4.6 in advanced software engineering at the same $5/$25 pricing, a new xhigh effort level, /ultrareview, higher-resolution vision and the first cyber safeguards. It is now a legacy model, succeeded by Opus 4.8, Opus 5 and Opus 5.5.
- Cursor
Cursor vs Claude Code in 2026: Which AI Coding Tool Is Better?
Neither tool is terminal-only or editor-only any more: Cursor ships a CLI and Claude Code ships a desktop app, IDE extensions and a browser surface. Choose Cursor for inline tab autocomplete and cross-lab model choice, though OpenAI plans to end Cursor's direct model access on a proposed November 12; choose Claude Code for one agent across surfaces.
Frequently Asked Questions
What is Fast mode for Claude Opus 4.8?
Anthropic's Fast mode page describes Fast mode as delivering the same Opus-level model intelligence with significantly faster speeds. At launch it cited 2.5x faster output token speeds for Opus 4.8 compared to standard Opus 4.8. It is in research preview.
How do you enable Fast mode in Claude Code?
Run /fast in Claude Code. On Pro, Max, Team and Enterprise plans it bills only to usage credits, which must be turned on first, and Team and Enterprise need an Owner to enable it. Since v2.1.280 /fast picks Opus 5.5 by default; switch to Opus 4.8 with /model to run it there.
How do you get Fast mode on the Claude Platform API?
Anthropic says API access is a limited research preview. Access requires contacting your account manager, and teams without an account manager can join the Fast mode waitlist by completing the form linked from the Fast mode page.
Does Fast mode change Opus 4.8's quality or cost?
Quality stays the same: fast mode is the same model at higher speed. At launch Anthropic made it three times cheaper than before, and the Claude Code docs price it at $10/$50 per million input/output tokens on Opus 4.8, above standard Opus. Fast mode on the newer Opus 5.5 costs $8/$40.