专题
返回时间轴
2026-07-16

Moonshot AI releases Kimi K3 — the largest open-source model ever at 2.8 trillion parameters

Capability Breakthrough Commerce & Industry

事件摘要

Kimi K3 (Jul 2026) — Moonshot AI's 2.8-trillion-parameter open-source MoE model with 1M-token context, native vision, and Kimi Delta Attention architecture. The largest open-source model ever released, rivaling top US frontier systems.

影响评估

  • Paradigm Shift +1 · Medium-term

    At 2.8 trillion parameters, Kimi K3 is the largest open-source model ever released, demonstrating that the scale ceiling for open-weight models continues to rise and that Chinese labs can produce competitive frontier models despite US export controls.

    Affected Groups: AI researchers, open-source community, AI developers

  • Economic Disruption +1 · Medium-term

    Kimi K3's open-source release continues the downward pressure on AI pricing, offering a no-cost alternative to proprietary APIs for organizations capable of self-hosting.

    Affected Groups: enterprises, startups, developers

共识度与来源

重要度 L1
分类 Capability Breakthrough / Commerce & Industry
共识度 Broad Consensus
影响指数 5/10

关联事件

← Origins
2025-01-20 L2

DeepSeek releases R1, an open-source reasoning model that shocks global markets

On January 20, 2025, Chinese AI lab DeepSeek released R1, an open-source reasoning model matching OpenAI's o1 on key benchmarks at a fraction of the training cost. Using Mixture-of-Experts architecture and Group Relative Policy Optimization, R1 achieved o1-level reasoning for approximately 1/20th of the training compute. On January 27, the release triggered a global stock market selloff that erased $593 billion from Nvidia's market cap — the largest single-day value loss in history — and DeepSeek became the most-downloaded free app on the US iOS App Store.

2026-04-24 L1

DeepSeek releases V4, an open-weight 1.6 trillion parameter model that undercuts all competitors on price

On April 24, 2026, DeepSeek released V4, its most powerful open-weight model family — with V4-Pro at 1.6 trillion total parameters (49B active) and V4-Flash at 284B (13B active), both under the MIT license with a 1-million token context window. Priced at $0.87 per million output tokens, V4 undercut GPT-5.5, Claude Opus 4.7, and Gemini 3.1 Pro by 5-10x. The release demonstrated that the open-weight ecosystem had permanently reshaped AI economics.

2026-06-13 L2

GLM-5.2 Open Source: Chinese AI Model Matches Frontier Closed-Source Performance

On June 13, 2026, Z.ai (formerly Zhipu AI) open-sourced GLM-5.2, a 744-billion-parameter Mixture-of-Experts model under the MIT license with a 1-million-token context window. The model achieved frontier-level coding and agentic performance — competitive with GPT-5.5 at roughly one-sixth the inference cost — marking a milestone where open-source AI meaningfully caught up to proprietary frontier models. The release was widely seen as a direct response to the US export ban on Anthropic's Fable 5, demonstrating that restrictive policies accelerate, rather than prevent, global AI capability diffusion.

Local Lineage Moonshot AI releases... 2026-07-16 · L1 DeepSeek releases ... 2025-01-20 · L2 DeepSeek releases ... 2026-04-24 · L1 GLM-5.2 Open Sourc... 2026-06-13 · L2