By
–
“Hits hard”????? Don't talk nonsense, @01net. The Sonnet 5 release makes absolutely NO SENSE. It's cheaper per request but needs many more requests to complete a task, so it's much more expensive than Opus 4.8 for the same task? So Opus [Traduit de FR vers EN]
By
–
AI Layoffs Are Here, But They Don’t Mean What You Think #Tech #layoffs are increasingly being linked to #AI, but the real story is more complex than #machines simply replacing humans. As companies pour billions into AI #infrastructure, #cloudcomputing and specialist #skills,
By
–
𝗧𝗵𝗶𝘀 𝗶𝘀 𝘄𝗵𝘆 𝗔𝗜 𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻𝘀 𝗯𝗿𝗲𝗮𝗸 𝗽𝗲𝗼𝗽𝗹𝗲’𝘀 𝗯𝗿𝗮𝗶𝗻𝘀. I’m skeptical of AI valuations that price every model company as if it will own the future. But I also believe the market is seeing something real. This comparison feels absurd because one
By
–
Anthropic seriously messed up with Claude Sonnet 5. The token efficiency is so bad it ends up being more expensive than Claude Opus 4.8. WTF are they doing?
By
–
Notes (and a Pelican) on Claude Sonnet 5 – the new tokenizer makes it ~1.4x more expensive for English, ~1.33x more expensive for Spanish but roughly the same price for Simplified Mandarin
By
–
For the RL chapters, I think you should be able to bet away with $10-$20 in cloud compute (in total) for an A100 or L40S.
By
–
Great work. But I would be cautious about interpreting employment growth among AI adopters as evidence against broader labor-market displacement. Firms that adopt AI may grow by gaining market share from non-adopters, so employment can rise among adopters even as exposed
By
–
Anthropic hates Opensource AI because a model like GLM 5.2 being available for feee to download made their $1 Trillion valuation sound like a joke (which it is btw)
By
–
If you use LLM-as-judge, this one is for you.
— Akshay 🚀 (@akshay_pachaar) 30 juin 2026
(bookmark it)
Most teams validate their agent's outputs by calling a frontier model as the judge. It works, until it doesn't.
Three problems stack up fast:
→ Cost: you're hitting a frontier API on every turn, every tool call,… pic.twitter.com/oLqPYKb0sD
If you use LLM-as-judge, this one is for you. (bookmark it) Most teams validate their agent's outputs by calling a frontier model as the judge. It works, until it doesn't. Three problems stack up fast: → Cost: you're hitting a frontier API on every turn, every tool call,
