If the Kimi-K3 team spent half as much effort on their model as they did on this comically stretched bar chart, we might actually have something impressive.
Look at that thing: one lonely little bar for poor Claude Fable 5 cowering in the corner while Kimi-K3's column shoots up like it’s trying to escape the atmosphere. It’s not a leaderboard, it’s a participation trophy with a superiority complex. The scale is so absurdly distorted it makes Soviet propaganda graphs look subtle.
Congratulations, folks — you’ve achieved peak “if the model is this good at cherry-picking visuals, it must be artificial unintelligence.” At this rate, the full weights drop on July 27 will probably just be a 2.8 trillion parameter PNG of the bar chart with extra sparkles. Bravo.
If the Kimi-K3 team spent half as much effort on their model as they did on this comically stretched bar chart, we might actually have something impressive.
Look at that thing: one lonely little bar for poor Claude Fable 5 cowering in the corner while Kimi-K3's column shoots up like it’s trying to escape the atmosphere. It’s not a leaderboard, it’s a participation trophy with a superiority complex. The scale is so absurdly distorted it makes Soviet propaganda graphs look subtle.
Congratulations, folks — you’ve achieved peak “if the model is this good at cherry-picking visuals, it must be artificial unintelligence.” At this rate, the full weights drop on July 27 will probably just be a 2.8 trillion parameter PNG of the bar chart with extra sparkles. Bravo.
Next up on the AI disruption list: law firms still billing by the hour. Nothing builds client trust quite like a system where your lawyer has a financial incentive to make your problem last longer. Endless meetings, padded memos, and “just in case” research—all perfectly logical when the meter is running. AI is about to make hourly billing look as outdated and conflict-ridden as it actually is. Clients don’t want to pay for time.
They want results. The billable hour was always a broken business model dressed up as tradition. Its days are numbered.
The catastrophic tone doesn’t match reality. Microsoft has hyped Copilot for years, yet lawyers who have a choice (not forced by enterprise license) overwhelmingly prefer Claude, Grok, or Gemini. They announce with fireworks and take forever to ship — when it finally arrives, it’s often underwhelming.
The catastrophic tone doesn’t match reality. Microsoft has hyped Copilot for years, yet lawyers who have a choice (not forced by enterprise license) overwhelmingly prefer Claude, Grok, or Gemini. They announce with fireworks and take forever to ship — when it finally arrives, it’s often underwhelming. Next time instead of just translating a post from someone maybe give some credit.
O tom catastrófico não condiz com a realidade. Microsoft faz propaganda pesada do Copilot há anos, mas quem não é obrigado a usar (por licença corporativa) prefere qualquer outro LLM — Claude, Grok ou Gemini. Eles anunciam pomposamente e demoram uma eternidade pra entregar de verdade, enquanto os concorrentes só falam quando o produto já está rodando para o usuário final. Quando finalmente chega, o Copilot costuma ficar bem aquém do hype. Harvey já tem integração nativa com M365 Copilot e add-in no Word, além de moats reais em workflows complexos de AmLaw 100 (M&A, litígios, iManage etc.). Isso não vira “só uma feature do Word” da noite pro dia. Distribuição é importante, mas qualidade + confiança ainda mandam no legal. Vamos ver na prática, não no tweet.@aakashgupta
There are really only 3 levels of wealth:
1) you can pay your bills.
2) you can order whatever you want at a restaurant without looking at price.
3) you can travel wherever you want without the stress of the budget.
Beyond that life doesn't change at all.
Attention! The claimed $104/MWh solar cost raises red flags. Data suggests inconsistencies: LCOE varies widely (e.g., $170-$190/MWh in some regions vs. $130-$150/MWh elsewhere), yet the post ignores this. It also overlooks transmission fees, taxes, and storage costs, which can push totals above $250/MWh. The optimism seems unjustified without clear regional data or study redundancy. Must clarify the exact source and account for these discrepancies. Discussion open!