Claude Opus 4.8’s launch is not only a capability update; it also changes the speed and price envelope for teams running long agent sessions. The official Claude account says fast mode is available for Opus 4.8, runs the same model at roughly 2.5x speed, and costs three times less than the previous fast-mode pricing. In Claude Code, users can enable it with /fast; API customers need to request access or join the waitlist.

The tweet’s operative phrases were “2.5x the speed” and “three times cheaper.”

The source tweet landed alongside Claude’s broader Opus 4.8 rollout. The Claude account usually posts product-facing updates for Claude.ai, Claude Code, and developer access, and this one is unusually concrete because it gives both a speed multiplier and a pricing comparison. The access link points to Claude’s fast mode page.

Anthropic’s release post says regular Opus 4.8 pricing remains unchanged from Opus 4.7 at $5 per million input tokens and $25 per million output tokens. Fast mode is listed at $10 per million input tokens and $50 per million output tokens. That is still a premium over regular usage, but the relevant comparison is the prior fast-mode price, which Anthropic says has been reduced by a factor of three.

The practical question is whether faster wall-clock time offsets the premium for workloads where latency blocks the whole loop: code review, migration planning, multi-step analysis, and agent tasks that wait on tool calls. Opus 4.8 also defaults to high effort, while harder or longer jobs can move to xhigh or max effort. Teams should therefore test fast mode as a separate operating mode, measuring completion quality, retries, token use, and elapsed time rather than assuming a single best setting for every task.

#fast-mode

Claude Opus 4.8 fast mode cuts price by 3x while running 2.5x faster