AI News
AI community posts, news and release updates
Latest updates
- Rant
Double standard? Anthropic trains on the whole web, but draws the line at Kimi distilling?
Published Jul 22, 2026A sarcastic take: Anthropic can train on the entire internet without permission, but Kimi distilling from Claude is where they draw the line.
- Anthropic
- Kimi
- Distillation
- double standard
A developer finds that Agentic Workflow's cache keepalive mechanism drives costs up 8x, questioning its design and calling for optimization.
- cost
- Agentic Workflow
- cache
- keepalive
A traveler used ChatGPT to assist in suing a Norwegian airline from New York and won $4,760 in compensation.
- ChatGPT
- 法律
- 经验
- 旅行
- News
Unlimited AI tokens aren't unlimited after all as US Army burns through supply
Published Jul 22, 2026The US Army's rapid consumption of AI tokens has exposed the unsustainability of 'unlimited' token promises, highlighting cost issues in AI services.
- 成本
- AI token
- 美国陆军
- Discussion
Kimi K3 Was in Internal Evaluation Before Fable 5 Was Available Long Enough to Be Distilled
Published Jul 22, 2026The author points out that Kimi K3 was under internal evaluation from April/May, well before Fable 5 was available long enough to be plausibly distilled. Kimi already outperformed Fable on BrowserComp; if K3.1 beats Fable 5 on coding benchmarks, will people claim it distilled Fable 6? Clearly it's not just distillation.
- Distillation
- benchmarks
- Fable 5
- USG
- Kimi K3
- Tutorial
I Had 11 AI Subagents Test Every PPTX Skill for Claude Code. Here's What Works
Published Jul 22, 2026The author systematically tested Claude Code's ability to handle PPTX files using 11 AI subagents, summarizing what works and providing a practical guide for developers.
- Claude Code
- Testing
- PPTX
- AI subagents
The author argues that the frenzy around Kimi K3 shows cheap intelligence drives compute demand up, not down. K3 had to pause signups due to compute shortage, textbook Jevons Paradox: efficiency increases demand.
- scaling
- Compute Demand
- Jevons paradox
- Kimi K3
- AI efficiency
- News
Compute shortage worsens: Meta, OpenAI, Nvidia invest billions, while neocloud stocks dip
Published Jul 22, 2026A roundup of recent compute demand signals: Meta upsized its data center to $50B+, Kimi K3 paused signups, OpenAI raised infrastructure to $750B, Nvidia bought 9.3% of Nebius, SK Hynix memory sold out beyond 2030. Yet neocloud stocks are down 20-30%, presenting an opportunity.
- Nvidia
- data center
- Infrastructure
- Neocloud
- Hyperscaler
- Kimi K3
- compute shortage
- News
OpenAI tries consulting – charging boots-on-the-ground prices to deploy agents
Published Jul 22, 2026OpenAI has launched a consulting service to help clients deploy AI agents, charging prices comparable to hiring human experts, sparking debate on the value of AI consulting.
- OpenAI
- AI agents
- Pricing
- consulting
Treasury Secretary Scott Bessent hinted at sanctions, possibly using Kimi as a test case for new penalties. US-China AI talks in September are heating up.
- Kimi
- AI regulation
- Geopolitics
- US China
- sanctions