Moonshot AI launches Kimi K2.7-Code HighSpeed

// 97d agoMODEL RELEASE

Moonshot AI launches Kimi K2.7-Code HighSpeed

Moonshot AI has introduced a new high-speed mode for its open-source multimodal coding model, Kimi K2.7-Code, delivering up to 6× faster generation speeds. The update achieves up to 260 tokens per second on shorter-context tasks and is currently rolling out to Kimi Code Beta.

// ANALYSIS

High-speed inference is the key to making AI-driven software engineering feel truly interactive and seamless, and this update directly targets the latency bottleneck of complex coding tasks.

–**Inference Speedup:** Speeds of 180–260 tokens per second match or exceed dedicated inference hardware platforms, making real-time autocomplete and chat highly responsive.
–**Multimodal Support:** The model retains its multimodal capabilities, permitting developers to include visual UI layouts and diagrams in their prompts at speed.
–**Beta Deployment:** Integrating this variant directly into Kimi Code Beta facilitates immediate developer testing and real-world performance feedback.

// TAGS

kimikimi-k2.7-codemoonshot-aiai-codingcoding-modelllmdevtool

DISCOVERED

97d ago

2026-06-15

PUBLISHED

97d ago

2026-06-15

RELEVANCE

8/ 10

AUTHOR

Kimi_Moonshot

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

MODEL2h ago

Alibaba drops Qwen-Image-2.1 with RGBA transparency

Alibaba's Qwen team has officially launched Qwen-Image-2.1, a 7.1-billion parameter Diffusion Transformer (DiT) paired with a Qwen3-VL 8B text encoder that unifies text-to-image generation and image editing within a single architecture. A defining feature highlighted in community testing is its native RGBA support, enabling direct transparency generation and RGB-to-RGBA decomposition without relying on post-processing segmentation tools.

UPDATE3h ago

Classifier.dev launches a Teams feature and introduces the Laya classification model in beta.

Classifier.dev has released a new Teams feature that enables users to invite team members, share access, and track usage across their organization. Additionally, the platform introduced Laya, a fast classification model created by the inventor of RLCD, which is now available in beta to help developers build real-world applications.

LAUNCH3h ago

FreeUpscaler upscales images and videos locally using WebGPU

FreeUpscaler is a completely free tool that allows users to upscale both videos and images by 2x or 4x their original resolution. It distinguishes itself by running entirely locally in the user's browser using WebGPU, meaning no files are ever uploaded to a server, ensuring privacy and immediate access without installation.