Curated by
JD Audena
life’s better when it’s bold, curious, and a little chaotic | building & becoming always | in service of the future | fellow omnipotentialite and friend to @KyleHudson
stacklist.com/jdaudena
life’s better when it’s bold, curious, and a little chaotic | building & becoming always | in service of the future | fellow omnipotentialite and friend to @KyleHudson
More in Signals Served #020 — Intelligence Got Cheap in Public
See all 15 →Signals Served #020 — Intelligence Got Cheap in Public
thevcconcierge.com
Model Inference Pricing Explanation - Kimi API Platform
platform.kimi.ai
SOXX Enters Bear Market: Why Ed Yardeni Says Semiconductor Stocks Could Fall Another 12%
benzinga.com
More from JD Audena
See all stacks →New York Fintech Week Events 2025
7 cards · Finance
New York Fintech Week People
36 cards
Startup Roster
11 cards · Finance
China's 2.8-Trillion-Parameter Kimi K3 AI Model
Moonshot AI has released Kimi K3, a 2.8-trillion-parameter open-weight AI model that represents the largest of its kind, reportedly outperforming Claude Fable 5 on the Frontend Code Arena benchmark. The release highlights China's continued progress in AI development despite U.S. compute export restrictions.
Built for AI agentsACO · 9744 tokens
Summary
Moonshot AI has released Kimi K3, a 2.8-trillion-parameter open-weight AI model that represents the largest of its kind, reportedly outperforming Claude Fable 5 on the Frontend Code Arena benchmark. The release highlights China's continued progress in AI development despite U.S. compute export restrictions.
Tags
kimi-k3 · open-weight-model · moonshot-ai · large-language-model · china-ai · frontier-ai · benchmarks
Key entities
Moonshot AI (organization, 0.95) · Kimi K3 (technology, 0.98) · Claude Fable 5 (technology, 0.85) · open-weight model (concept, 0.9) · Frontend Code Arena benchmark (concept, 0.85) · China (location, 0.9) · U.S. compute limits (concept, 0.8) · Tom's Hardware (organization, 0.9)
Classification
notes · language en · status final
Provenance
claude-opus-4-6 via @stacklist/mcp-server@2.0.0, confidence 0.85, 21 Jul 2026