Anthropic has launched a fast mode for Claude Opus 4.6, marketed as Claude Fast Mode, delivering up to 2.5x faster responses while preserving quality, and it is available in a limited research preview through the Claude Developer Platform API and Claude Code.

Access is restricted during the initial testing phase: fast mode is available to users with extra usage enabled and is being tested by Anthropic’s internal teams before broader access, with an official X post announcing the feature.

Users can enable fast mode via the /fast command in Claude Code, and it is designed to speed up live debugging, rapid code iteration, and time-critical tasks, while standard mode remains recommended for longer autonomous runs or cost-sensitive workloads.

The Opus 4.6 update introduces improvements in planning and handling long-running tasks, enabling better tracking across large codebases and extensive documents with reduced context loss.

Early reports indicate fast mode costs roughly six times more than standard mode, reflecting the trade-off between speed and expense.

Anthropic emphasizes that fast mode aims to maintain Opus-level intelligence and quality despite higher costs, targeting speed for critical tasks.

The rollout is being treated as an experiment; feedback will be gathered before any wider public release, with a research preview ongoing.

Fast mode is not yet available through major cloud providers like Amazon Bedrock, Google Vertex AI, or Microsoft Azure Foundry; a waiting list exists for interested developers and API access may expand later.

Beyond coding, Claude Opus 4.6 is positioned as a general-purpose work model for tasks such as financial analysis, research, and handling documents, spreadsheets, and presentations.

Pricing varies by input and output tokens, with fast mode potentially costing up to six times standard rates; a 50% introductory discount is offered until mid-February.

Internal teams have already been using the faster version, and access is being gradually extended to Claude Code users and API participants as part of the performance evaluation.

Fast mode is intended for urgent, high-stakes projects, while standard mode is better suited for longer, cost-sensitive workloads.