Anthropic launches fast mode for Claude Opus 5.5

Anthropic has made the fast mode of Claude Opus 5.5 accessible to all developers. The previous waitlist has been removed, and a separate application is no longer required. In this mode, Opus 5.5 generates answers up to 2.5 times faster than before, but the costs per word have also doubled.

On October 15, 2026, Anthropic announced via the developer X account that the speed mode is now available. Previously, this mode was an experimental feature for which you had to sign up. The documentation of the Claude Platform confirms that access to Opus 5.5 in this mode is no longer restricted.

The speed increase comes from the same model, Opus 5.5, which now operates more efficiently. However, the quality and capabilities of the model remain unchanged. The difference in speed is particularly noticeable with longer answers, such as complex code. Developers can easily activate the mode by adding a line 'speed: fast' in their request.

Nevertheless, costs have risen; where using Opus 5.5 normally costs $4 per million input tokens, these costs in fast mode are $8. This means that the total cost for long texts is significantly higher. For older models, the fast mode remains in testing phase for now and requires a sign-up. While Microsoft Foundry already supports the fast mode, Google Cloud has yet to announce its implementation.

The recommended options for cost-conscious developers are the cheaper alternatives within the Claude family, such as Claude Haiku 5.5, which is on average 75 percent cheaper. For Dutch developers working with Claude Code, the fast mode remains subject to availability, with price and availability possibly varying. Anthropic positions speed as a paid service; this offers advantages for those looking to optimize timeframes with an AI assistant, but for large projects, the standard speed remains the more cost-effective choice.

Read the full article from AI Insider.