Yes. It’s not that we’ve discovered some magic bullet, but rather that JAX, or at least the open source version of it, is mostly optimized for small to medium-sized training runs on Google TPUs, whereas we need to massive training runs on Nvidia GPUs. Pipeline parallelism is
RESEARCH
-
AI Model Opus 4.8 Calibration and User Feedback Request
By
–
We put a lot of work into calibrating thinking effort for Opus 4.8. As you're trying out the model, if you do run into any examples of it still over/under thinking, please flag it to us!
-
ByteDance Open-Sources Capable Multimodal AI Model BAGEL
By
–
ByteDance just open-sourced one of the most capable multimodal models out there.
— Chubby♨️ (@kimmonismus) 28 mai 2026
BAGEL does image generation, editing, style transfer, and visual understanding – all in a single 7B parameter model. Apache 2.0 licensed!
One model. No switching between specialized tools.
Amazing https://t.co/r4jj4QHTzMByteDance just open-sourced one of the most capable multimodal models out there. BAGEL does image generation, editing, style transfer, and visual understanding – all in a single 7B parameter model. Apache 2.0 licensed! One model. No switching between specialized tools.
Amazing -
Claude Opus 4.8 Released: Stronger Coding Model with Benchmarks
By
–
Claude Opus 4.8 is out today. It's our strongest coding model yet: up on SWE-bench Pro (from 64.3 to 69.2) and noticeably more honest about its own work. It tells you when it's unsure and catches its own bugs instead of declaring victory early. Same price as 4.7.
-
Anthropic Releases Claude Opus 4.8: Faster, Cheaper, and Smarter
By
–
ANTHROPIC ACABA DE LANZAR CLAUDE OPUS 4.8
— Nico (@nicos_ai) 28 mai 2026
Y puede que sea su mejor modelo hasta la fecha:
→ 2.5x más rápido y 3x más barato con modo /fast
→ Trabaja SOLO como un ingeniero senior, sin que lo supervises
→ Lanza cientos de subagentes en paralelo para completar tareas complejas… https://t.co/QSIQ1rvfpo pic.twitter.com/2cNoRMTRDgANTHROPIC JUST RELEASED CLAUDE OPUS 4.8 And it might just be their best model to date: → 2.5x faster and 3x cheaper with /fast mode
→ Works ONLY like a senior engineer, no supervision needed
→ Launches hundreds of subagents in parallel to handle complex tasks on its own And -
User compares AI model 4.8 to 4.6, highlighting collaboration strengths
By
–
„4.8 understands nuances better, feels much more natural to talk to, and is overall a stronger collaborator on everything from coding to knowledge work.“ So big. Is 4.8 being our good old friend 4.6 just better?? Testing time
-
AI+Science Conference Highlights: AI Transforming Scientific Discovery
By
–
Missed our AI+Science conference? Read the highlights, go deeper with the conference notebook, view photos from the day, and see a playlist of the recordings –– all in one spot: https://
hai.stanford.edu/news/how-ai-is
-transforming-scientific-discovery-while-keeping-humans-at-the-center
… -
Opus 4.8 AI Model Release: Agentic Coding Boost, Faster, Cheaper Modes
By
–
Opus 4.8 is live. Benchmarks especially significant jump in Agentic coding, but more important: „Fast mode is available for Opus 4.8. It's the same model at roughly 2.5x the speed, and we've made it three times cheaper than before.“
-
Anthropic Releases Opus 4.8 Model with Benchmark Results
By
–
BREAKING:
— Dan Shipper 📧 (@danshipper) 28 mai 2026
Anthropic just dropped Opus 4.8—and it is a MONSTER
We've been testing for about a week @every and our verdict is they could've just called it Opus 5, it's that good.
Here's our vibe check:
– Beats GPT-5.5 on Senior Engineer bench. On our toughest benchmark Opus… pic.twitter.com/Qsyn0AShn8BREAKING: Anthropic just dropped Opus 4.8—and it is a MONSTER We've been testing for about a week @every and our verdict is they could've just called it Opus 5, it's that good. Here's our vibe check: – Beats GPT-5.5 on Senior Engineer bench. On our toughest benchmark Opus
-
Release of Opus 4.8: Improved AI Model for Coding and Knowledge Work
By
–
Excited to release Opus 4.8 today! We heard your feedback on 4.7 and have made many fixes for 4.8. 4.8 understands nuances better, feels much more natural to talk to, and is overall a stronger collaborator on everything from coding to knowledge work.