PulseAugur
EN
LIVE 18:57:31

Moonshot's Kimi K2.7 Code debuts on Fireworks with enhanced agent efficiency

Fireworks AI is launching support for Moonshot's Kimi K2.7 Code, a new coding model designed for agentic tasks. This model boasts a 1T parameter architecture with 32B active parameters per token and a 256K context window, showing improved efficiency by using 30% fewer reasoning tokens than its predecessor while achieving higher scores on coding benchmarks. The company highlights that this token efficiency is crucial for agentic workflows, where costs can escalate rapidly due to repeated token usage in reasoning chains. AI

IMPACT Enhanced agent efficiency and reduced task costs for AI developers, particularly in long-horizon coding tasks.

RANK_REASON New model release from a frontier lab (Moonshot) with detailed technical specifications and performance metrics. [lever_c_demoted from frontier_release: ic=1 ai=1.0]

Read on Fireworks AI blog →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Moonshot's Kimi K2.7 Code debuts on Fireworks with enhanced agent efficiency

COVERAGE [1]

  1. Fireworks AI blog TIER_1 English(EN) ·

    Kimi K2.7 Code on Fireworks: Better Agents, Lower Cost per Task, Available Day

    Fireworks is launching Day-0 support for Kimi K2.7 Code, the latest model in Moonshot’s K2 coding series. Designed for long-horizon agentic coding, K2.7 Code leverages the established 1T parameter, 256K context window architecture while delivering a significant efficiency breakth…