the little whale is beached… yes, yes, more suspense please
the little whale is beached… yes, yes, more suspense pleaseamrit: Deepseek fell off or something
Latest
Trellis.cpp now produces high quality assets
Some of you might remember that I posted some time ago about the GGML-ported asset production pipeline. A key elelent of that was the TRELLIS.2 port…
Ask HN: Any AWS billing issues known? Amazon forecast of 3 billion dollars
I receive an AWS Budgets alert that my budget is exceeding the alert threshold. Threshold is 5$. Forecasted amount is listed as $3,005,575,870.47. (Yepp, right, that’s…
RT Ankith 🐋/acc: If you remove mathematics and language, kimi is #3 and ahead of Fable If you remove only mathematics, kimi is still #3 but only in…
RT Ankith 🐋/accIf you remove mathematics and language, kimi is #3 and ahead of FableIf you remove only mathematics, kimi is still #3 but only inferior…
Chinese degens curse Wenfeng for parting them from their money and putting it to better use (AGI)
Chinese degens curse Wenfeng for parting them from their money and putting it to better use (AGI)尚月: @teortaxesTex Interestingly, now Liang Wenfeng's image in the Chinese…
Just got an AWS billing alert projecting my monthly cost at $140B
This has to be a bug. My usual monthly S3 bill is only $2–3. Has anyone else received a strange billing estimate today?https://prnt.sc/UqjcYD3RSQrS Comments URL: https://news.ycombinator.com/item?id=48945606…
b10057
sycl: fix row calculation when K_QUANTS_PER_ITERATION is 1 (#25690) sycl: fix incorrect row calculation when K_QUANTS_PER_ITERATION=1 Signed-off-by: Todd Malsbary todd.malsbary@intel.com sycl: use K_QUANTS_PER_ITERATION for non-reordered Q5_K…
RT X Freeze: Turns out Elon was completely right The coding moat is disappearing in real time
RT X FreezeTurns out Elon was completely rightThe coding moat is disappearing in real time
RT Hugging Apps: SenseNova U1 is the best open image model you maybe didn't heard of yet Unified language-image (like Nano Banana 2 / GPT-Image-2) tha…
RT Hugging AppsSenseNova U1 is the best open image model you maybe didn't heard of yetUnified language-image (like Nano Banana 2 / GPT-Image-2) that does precise…
Joseph Needham would have been so happy today
Joseph Needham would have been so happy today
"definitely ahead of Opus 4.8"
"definitely ahead of Opus 4.8"Elliot Arledge: http://x.com/i/article/2077990815679361024
would be really sad if Chinese models were only frontier on SWE slop (excluding cyber) and "general reasoning" (excluding biology). But oh well! more …
would be really sad if Chinese models were only frontier on SWE slop (excluding cyber) and "general reasoning" (excluding biology). But oh well! more business for…
LongStraw: Long-Context RL Beyond 2M Tokens under a Fixed GPU Budget
A growing gap separates inference context lengths from RL post-training: inference systems are approaching million-token contexts, while post-training workloads often remain at 256K tokens or below…
A scorecard for the AI age
Sarah Friar, CFO of OpenaAI, introduces a practical AI scorecard to measure ROI through useful work, cost per successful task, dependability, and return on compute.
SolutionsIntroducing physics AI at Mistral: the foundation for engineering acceleration.A new class of AI models that predict the behavior of physical systems, powering the engineers and hardware products of tomorrow.May 27, 2026By Mistral
SolutionsIntroducing physics AI at Mistral: the foundation for engineering acceleration.A new class of AI models that predict the behavior of physical systems, powering the engineers and…
interestingly, I was wrong on who'll do this. I expected DeepSeek to go for 16 activated experts out of a large pool. Instead, they did 7/384, and Kim…
interestingly, I was wrong on who'll do this. I expected DeepSeek to go for 16 activated experts out of a large pool. Instead, they did 7/384,…
ProductConnect the dots: Build with built-in and custom MCPs in StudioConnect enterprise data to your AI applications with reusable connectors, direct tool calling, and human-in-the-loop approval controls.May 22, 2026By Mistral AI
ProductConnect the dots: Build with built-in and custom MCPs in StudioConnect enterprise data to your AI applications with reusable connectors, direct tool calling, and human-in-the-loop approval…
I Owe My Life to the Commodore 64
Article URL: https://www.goto10retro.com/p/i-owe-my-life-to-the-commodore-64 Comments URL: https://news.ycombinator.com/item?id=48945260 Points: 17 # Comments: 4
it's so peak… they should have contracted with any neocloud this is intolerable
it's so peak…they should have contracted with any neocloudthis is intolerable
AWS: Inaccurate Estimated Billing Data – $1.7 BILLION
URL already posted: https://health.aws.amazon.com/health/statusI've got an estimated bill for $1.7 BILLION over this month. Normal usage is < $5.Obvs have created an urgent AWS support ticket.…
> oai and a are still leading the real frontier, glm can match opus with a <1T model, dpsk has incredibly high inference efficiency, we have very muc…
> oai and a are still leading the real frontier, glm can match opus with a
RT Elliot Arledge: http://x.com/i/article/2077990815679361024
RT Elliot Arledgehttp://x.com/i/article/2077990815679361024
Honesty Is a Competitive Advantage
https://preview.redd.it/8ank6w5gardh1.png?width=1745&format=png&auto=webp&s=3c6c25261feaa6ecfe82d23ef627dfb43181e6e9 Huge respect to the Kimi K3 team for putting this sentence right at the beginning of their launch post. Instead of pretending to be the…
The risk of weather data sabotage is rising
Every morning, airline dispatchers, grid operators, and farmers around the world make decisions based on the same thing: a weather forecast. While these forecasts are something…