According to AIbase, Anthropic has recently introduced a new "Fast Mode" for its flagship model Claude Opus 4.6. This mode sacrifices some cost-effectiveness to achieve extreme response efficiency, with a speed that is 2.5 times faster than the standard mode.

Although the response speed is significantly improved at the same quality level, the pricing of "Fast Mode" is extremely high. According to the pricing standards announced by Anthropic, the cost of this mode can be as high as 6 times that of the standard mode. The specific pricing is as follows in the table:
| Item | Standard Mode (per million tokens) | Fast Mode (per million tokens) |
| Input (≤200K Tokens) | $5 | $30 |
| Input (>200K Tokens) | $10 | $60 |
| Output (≤200K Tokens) | $25 | $150 |
| Output (>200K Tokens) | $37.50 | $225 |
Applicable Scenarios and Limitations
Anthropic states that Fast Mode is designed for real-time debugging, rapid code iteration, and time-sensitive tasks. For longer-term autonomous operations, batch processing, CI/CD pipelines, or cost-sensitive tasks, the company still recommends using the standard mode.
Activation Method: Users can enable it in Claude Code by entering
/fast, or use it in integrated environments such as Cursor, GitHub Copilot, Figma, and Windsurf.Platform Restrictions: This mode is currently not available through Amazon Bedrock, Google Vertex AI, or Microsoft Azure Foundry. Anthropic says they will expand API access in the future, and developers can now apply to join the waiting list.
Time-limited Offer: As of February 16, this mode offers a 50% discount (i.e., the current price is half of the final price).


