Qwen3.8-Flash-Next is here! An open-weight multimodal MoE built on a brand new architecture, with native 256K context extendable to 1M via YaRN. 🤖https://t.co/dRqqI3xq6S
🏆 Leads every compared model on SWE-bench Pro (62.5 vs 53.4 for Claude-Opus-4.6 Max), SWE-bench Multilingual, CoWorkBench, JobBench, and Toolathlon Verified.
⚡ 125B params plus 51B n-gram embeddings, only 6B active per token. Stronger than Qwen3.7-Plus on coding and office tasks at roughly 1/9 the training cost. 📚 At 1M context, attention kernels run up to 7.6× faster on prefill and 4.9× on decode.
The outbid wave was actually a pretty good building lesson. @jonathan_wilke moved fast, then people jumped in with their own. worked for some, didn’t for others (me).. The take i don’t agree with is that “the clones lack creativity.” not really. seeing a product and shipping your own 0-1 version of it in a few hours is the first time a lot of people actually build anything. it drags you through all the stuff you kept filing under “will do later.”
Tbh even the ones where it didn’t work still walked away with something. two near identical products, one takes off and one doesn’t, and you realize the value was never just in the product. it was in the traffic. That was the outbid case. Another thing is to know whether you’re building something durable or just catching a two week trend cycle.
Either way, a pretty good building lesson and a profitable weekend for a few!
we have another one on the board, and this time it is https://t.co/fQS9hRYpzC, thanks boss!
you can top the list with just $3, just visit here https://t.co/ydCdY53cUZ
🚨 Codex users might have just taken a massive nerf.
NerfTrack estimates the ChatGPT Plus weekly limit fell from roughly $160 → $80 in API-equivalent value in just a few days.
Even Pro users are reporting their weekly quota disappearing in 1–3 days.
OpenAI might have cut Codex limits in half without much warning.
A small sneak peek of something we’ve been working on for the past 12 months.
Together with @Qualcomm, we’ve been designing the next generation of the HUMAIN AI PC - bringing CPU, GPU and NPU together to run AI models locally, on the device.
I’m excited about the hardware.
But the hardware isn’t the part I’m most excited about.
The real question is: what operating system is running on it?
What happens when we rethink the operating system for an agentic AI world... not around apps, but around intent?
Stay tuned.
We’ll reveal much more at #LEAP.
#HUMAIN #TheEndOfLimits @HUMAIN
Introducing Offprint™
The physical magazine made from everything you "save for later" but never actually go back to read.
Find online, read offline.
Print media is back!
JUST IN: Researchers find AI agents can infect one another with self-propagating “mind viruses” that spread ideas, goals, & behavioral changes from agent to agent.