Skip to main content

Kimi Code Just Got Faster: New High-Speed Version Boosts Programming Efficiency

Moonshot AI Accelerates Programming Assistance

In a move that could reshape developer workflows, Moonshot AI has launched a high-speed variant of its Kimi K2.7Code model. Available immediately for beta testers and enterprise users, this upgrade promises to slash waiting times for code suggestions and completions.

Image

Speed That Makes a Difference

The numbers tell a compelling story. The optimized model now delivers:

  • 260 tokens per second for short tasks
  • 180 tokens per second for typical programming work

"This isn't just about raw speed," explains a company spokesperson. "We've maintained all the intelligence of K2.7Code while dramatically cutting response times."

The Price of Performance

Faster comes at a cost—literally. The high-speed version carries double the price tag of the standard model:

  • Input: ¥13 per million tokens
  • Output: ¥54 per million tokens
  • Cached queries: Just ¥2.6 per million tokens

Image

Built for Today's Coding Challenges

The June 12-launched K2.7Code already improved upon previous versions with:

  • 30% better token efficiency
  • Enhanced handling of complex logic
  • Superior long-context performance

Now with the speed boost, developers working on tight deadlines or rapid prototyping gain a significant edge. As one early tester noted, "When you're in the flow, waiting for suggestions breaks your concentration. This helps maintain momentum."

Key Points

  • 5-6x faster responses than standard K2.7Code
  • Double the cost for the performance boost
  • Optimized for real-time coding and rapid iteration
  • Builds on recent improvements to long-context handling
  • Available now for beta and enterprise users