Kimi K2 Turbo-Preview Delivers Breakthrough Speed

The Kimi K2 Turbo-Preview, a high-performance variant of the popular AI model, has officially launched with 40 tokens per second output speed—a fourfold increase over its predecessor's 10 tokens per second. This release maintains the original Kimi-k2's parameter configuration while dramatically enhancing processing efficiency.

Limited-Time Pricing Promotion

To celebrate the launch, developers are offering a 50% discount through September 1:

• Input (cache hit): ¥2.00/million tokens
• Input (cache miss): ¥8.00/million tokens
• Output: ¥32.00/million tokens

This strategic pricing positions Kimi K2 Turbo-Preview as a highly competitive option in the AI inference market during the promotional period.

Future Development Roadmap

The development team emphasized this release marks just the beginning of their optimization efforts. "We're committed to pushing speed boundaries further while maintaining quality," stated a company representative. Future updates may include additional performance enhancements and feature expansions.

Key Points:

• 4x speed boost from original model (10 → 40 tokens/sec)
• Identical parameters ensure consistency with existing implementations
• Promotional pricing available until September 1, 2025
• Ongoing optimizations planned for future releases