Many users are impressed by Moonshot AI open-sourcing Kimi K3 for frontier coding benchmarks at just $15-25M because the low cost and efficiency stand out against typical expensive frontier efforts.
Based on 14 visible X reactions from 24 accounts; directional sample.
Ask a question below.
Published answers will appear here.
The $20 million model sits at the EU systemic risk compute threshold.
@EMostaque @Kimi_Moonshot Did it actually perform with actual testing? My experience so far with Chineese models is usally bench maxing, when I use them in existing mature system, instead of Claude/Codex/Grok (lately) they are not as good.
@EMostaque @Kimi_Moonshot i'm legitimately blown away, foothills of the singularity indeed.
@EMostaque WOOO, congrats to Kimi K3 on being a true frontier model!.
@EMostaque @Kimi_Moonshot Super impressive.
Worth noting the EU AI act put the threshold for AI models with systemic risk at 1e25 flops (!) US executive order put it at 1e26 flops of training compute.. Oddly prescient
Kimi K3 seems fine-tuned to dominate the UI coding space. What other domain can benefit from a $18m training run?
Is this like 1/8th the cost of a Hollywood blockbuster film?
Many users are impressed by Moonshot AI open-sourcing Kimi K3 for frontier coding benchmarks at just $15-25M because the low cost and efficiency stand out against typical expensive frontier efforts.
Based on 14 visible X reactions from 24 accounts; directional sample.
Ask a question below.
Published answers will appear here.