February 7, 2026
Speed thrills, bills kill
Speed up responses with fast mode
Same answers, faster replies—if you’ll pay the rush fee
TLDR: Anthropic’s Fast mode makes the same AI reply quicker but costs more per chunk of text and bills extra usage from the first token. Comments are split: price shock and demands for real speed numbers, plus fears that regular mode gets slower to push people into paying.
Anthropic just dropped “Fast mode” for its Opus 4.6 assistant—same brain, quicker mouth, bigger bill. Flip it on with /fast and watch a little lightning bolt appear as replies speed up. It’s the same model, just tuned for speed, not savings, with a limited-time 50% discount until Feb 16. Tokens (tiny chunks of text) get pricey here—input and especially output—and that’s where the community went full popcorn. Full details live on the Fast mode page.
The loudest reaction? Sticker shock. speedping saw “$30/150 MTok” and said “Umm no thank you,” while thehamkercat labeled the output price “insane.” Meanwhile, simonw asked the real question: How much faster, exactly? Without hard numbers, folks feel like they’re buying a mystery speed-boost. 1123581321 pointed out it’s “billed directly to extra usage… from the first token,” turning the feature into a “cover charge” for speed and invoking the $50 extra-usage credit as a test drive. Then came the drama: krm01 worried this creates a pay-to-go-fast world where standard mode gets slowed to nudge you into the upgrade. And that “same quality” promise met side-eye when the docs also mention “less thinking time” possibly hurting complex tasks—cue memes about a turbo mushroom that sometimes skips the corners. Devs on tight deadlines are excited; batch-job folks rolled their eyes. The vibe: speedrun mode for your code—just don’t look at the meter.
Key Points
- •Fast mode configures Opus 4.6 to prioritize speed over cost, keeping model quality and capabilities the same.
- •Toggle fast mode via /fast in Claude Code CLI and VS Code extension; it persists across sessions and auto-switches to Opus 4.6.
- •Pricing starts at $30/MTok input and $150/MTok output (<200K tokens), and $60/$225 (>200K tokens), with a 50% discount until Feb 16 (PT).
- •Fast mode requires extra usage enabled and admin approval for Teams/Enterprise; it is not available on Amazon Bedrock, Google Vertex AI, or Azure Foundry.
- •Fast mode has separate rate limits and falls back to standard Opus 4.6 during cooldown; enabling mid-conversation incurs full uncached input costs for the entire context.