Moonshot’s Kimi K3 arrives as a 2.8T open-weight model with 1M context
bengoertzel · x · 2026-07-21
Moonshot AI released Kimi K3 on July 16 as a 2.8T-parameter MoE model with a 1M-token context window, native multimodal input, reasoning enabled by default, and a plan to publish full weights under a permissive Modified MIT license by July 27.
Launch-week benchmarks reportedly put it just behind Claude Fable and GPT-5.6-sol and ahead of other models on several metrics, including long-horizon coding and agentic evaluations. The post also notes speculation about distillation from proprietary Western models, and says markets briefly sold off on the launch before partially recovering as analysts argued the cloud/inference demand still benefits hyperscalers.
Related event: Moonshot Releases 2.8T Open-Weights Model Kimi K3(14 posts)→
More from Venture
- Investor argues Palantir-Nvidia partnership should slash Anthropic's IPO valuation — pdamodaran · 2026-09-11
- Moonshot's annualized revenue jumped from $300M to $1B in two months after Kimi K3 — Hesamation · 2026-09-11
- Mid-market companies' AI SEO bottleneck is ops execution, not strategy, says SEO practitioner — gaganghotra_ · 2026-09-11
- A YouTuber with 1.5M followers paid this indie maker for a consulting call — tibo_maker · 2026-09-11
- 71% of people have never used generative AI — the bubble argument for microsaas — iamaliveix · 2026-09-11
- Glean grew from $100M to $300M ARR in roughly fifteen months — yogthinks · 2026-09-11