Kimi K3 Stuns the Tech World - Contrary Research
contraryresearch.substack.com · 1,290 words · saved by 1 readers
Research Rundown #193, plus new reports on Stoke Space and Aalyria
Chinese AI company Moonshot AI released Kimi K3 this week, a 2.8-trillion-parameter open-weight model that benchmarks ahead of Claude Opus 4.8 and GPT-5.5 on overall capability. It is the first Chinese open-weight model to achieve that milestone. It is also the largest open-weight model ever released, priced at $3 input / $15 output per million tokens, comparable in cost to Anthropic’s Sonnet tier. Its model weights will be published by July 27, at which point the benchmarks can be independently verified. There are at least two big implications of this news. First, regarding the US-China AI…
saved by
related reading
- Open-source Kimi K3 model is no reason for China panicsubstack.com
- Moonshot is Chinese But Its AI Models Are From Another Planetthealgorithmicbridge.com
- The Geopolitics of Open Weightsmbi-deepdives.com
- Who’s Afraid of Chinese Models? – Stratechery by Ben Thompsonstratechery.com
- Dario Amodei — On DeepSeek and Export Controlsdarioamodei.com
- AI 2027ai-2027.com
- Kimi K3 Tech Blog: Open Frontier Intelligencekimi.com
- Paper AI Tigersgleech.org
- Open Models Tack Toward the Frontiertomtunguz.com
- Dario Amodei — On DeepSeek and Export Controlsdarioamodei.com
- Google "We Have No Moat, And Neither Does OpenAI"semianalysis.com
- Our position on open-weights models \ Anthropicanthropic.com