Moonshot Prepares 3-Trillion Parameter Kimi K3

// 2h agoMODEL RELEASE

Moonshot Prepares 3-Trillion Parameter Kimi K3

According to a report by the Financial Times, Chinese AI startup Moonshot is poised to release its new Kimi K3 model as early as tonight. The model is rumored to possess between 2 and 3 trillion total parameters, which would make it the largest AI model released in China so far.

// ANALYSIS

Moonshot's rumored Kimi K3 release indicates that China's AI scaling wars are accelerating, even in the face of chip export restrictions.

–A 2-to-3 trillion parameter model would place Moonshot at the forefront of scale within China, challenging major tech players like Alibaba and Tencent.
–The massive parameter size suggests a Mixture-of-Experts (MoE) architecture is likely being used to keep inference latency and cost manageable.
–While parameter count is impressive, the actual performance gains and efficiency of Kimi K3 compared to global frontier models will be the key metric to watch.

// TAGS

moonshotkimi-k3chinallmmodel-releaseartificial-intelligence

DISCOVERED

2h ago

2026-07-16

PUBLISHED

2h ago

2026-07-16

RELEVANCE

8/ 10

AUTHOR

Kenjatina_og

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

OPEN SOURCE1h ago

Mindwalk visualizes AI agent sessions in 3D

Mindwalk is an open-source local tool that replays an AI coding agent's terminal session by illuminating the files it reads and edits on a 3D visualization of the repository. By scanning local projects and session logs, it renders a browser-based "night map" where files glow with specific colors (moss green for seen, moon white for read, warm amber for edited, and dark for unvisited), allowing developers to easily trace the agent's path, discover hallucination loops, and verify its overall pathfinding efficiency.

OPEN SOURCE1h ago

Clodex IDE launches open-source agentic sandbox

Clodex is an open-source, local-first agentic IDE designed to run autonomous AI tasks in isolated, user-approved environments. By treating engineering work as stateful tasks, it retains context across sessions, routes queries dynamically between models, and generates cryptographically signed evidence records for all operations.

OPEN SOURCE1h ago

Waggle optimizes multi-agent handoffs

Waggle is an open-source Rust library and MCP-native reference layer designed to streamline multi-agent workflows by passing compact, ~30-byte versioned reference tokens instead of massive context files during handoffs. Subagents resolve these tokens via the Model Context Protocol to retrieve only the specific data segments they need, reducing token bloat and enabling efficient context shaping and read attribution.