Moonshot AI released Kimi K2.7-Code on June 12, 2026, and on June 15 announced a HighSpeed Mode delivering up to six times faster token throughput, yet the Beijing-based lab has not submitted the model to any independent coding benchmark. The model is live on Hugging Face under a Modified MIT License and available on Moonshot's Kimi API at $0.95 per million input tokens and $4.00 per million output tokens, while Kimi Code starts at $19 per month. Moonshot claims the model can reduce reasoning token usage by roughly 30% compared to its predecessor, K2.6, but the core efficiency argument rests entirely on five proprietary benchmarks that Moonshot designed and ran itself. The HighSpeed variant promises throughput of around 180 tokens per second on median-length coding inputs and up to 260 tokens per second on shorter-context tasks. K2.7-Code is the fifth major model release in the Kimi series since July 2025, and Moonshot is now valued at more than $20 billion following a funding round that closed in May 2026. The model runs on a Mixture-of-Experts architecture with 384 specialist subnetworks, activating the top eight experts plus one shared expert per token, with an active parameter count of approximately 32 billion per token. The attention mechanism is Multi-head Latent Attention, originally developed by DeepSeek.