vivo ships four Blue LM models aimed at phones that actually finish tasks

Share
vivo ships four Blue LM models aimed at phones that actually finish tasks

The quiet part of the model race is happening on handsets, and vivo just laid out its full hand at its 2026 developer conference. Two of today's other stories point the same way: memory, not compute, is the industry's binding constraint, and the biggest brain map ever built is now a playground.

vivo rebuilt its Blue LM line around agents that act, not chat. At the company's 2026 developer conference on September 16, vivo announced four core models — Blue LM-RealTime, Blue LM-Nano, Blue LM-Flash and Blue LM-Pro — split across speech, on-device memory and cloud task execution. RealTime is an end-to-end speech model the company says holds up in noisy rooms, with accents, murmurs and mid-sentence language switching; Flash and Pro sit in the cloud and drive vivo's assistant across apps, devices and scenarios through an in-house agentic engine wired into the OS. vivo's pitch is that the system keeps learning from the corrections a user makes while a task runs. That is the same architecture bet every phone maker is now making: the handset vendor that owns the harness, not just the model, owns the assistant. With close to 600 million global users, vivo has more installed surface than most labs shipping agents have devices — the open question is whether a phone OEM can keep an agentic stack current at frontier speed.


Micron says memory, not silicon logic, sets AI's ceiling — and real new supply is years out. A Micron executive, Sumit Sadana, argued that memory bandwidth and capacity have become the key determinants of how far an AI system can be pushed, and that substantial new capacity will not come online until after 2028, according to reporting carried by Wall Street CN and Futu. That is the HBM bottleneck said out loud by one of only three firms that can relieve it: a 2028 horizon is a planning assumption every lab, hyperscaler and nation-state buyer now has to price in, and it doubles as a warning that the memory oligopoly keeps its pricing power far longer than the bear case allows.


The first complete map of an adult male fruit fly's brain is live, and modders immediately made it play Doom. Google Research and HHMI Janelia published the full wiring diagram of the male fly's brain and central nervous system — the largest connectome to date — and the map's open availability set off a wave of experiments in which people wired the virtual fly's motor outputs to game controllers, running Doom and trading Bitcoin, as 404 Media reported. The serious point underneath the stunts: with the female fly's map already published, researchers can now compare two complete brains of the same species, which turns connectomics from an atlas project into a test bench for how wiring produces behavior — the same substrate idea behind every "simulate the organism, skip the training" argument in AI.

If the phone is the new agent runtime, does it matter who has the best frontier model?

Read more

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kuaishou's Kling AI has spent two years becoming the video model people actually finish projects with. Its new flagship, Kling 4.0, went fully live on October 15 — and this round is less about prettier frames than about the two things that make a generated shot usable in a real edit: length and control. Kling 4.0 generates up to 30 seconds in a single native pass, outputs 10-bit HDR at 4K, 1080p and 21:9, and takes up to 15 reference assets in one task. That budget splits across ten images, fi

Brain organoids classified Braille at 83% — in a trio, not alone

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Microsoft finished pre-training its biggest in-house model yet — then used it to make a point about the company it bankrolls. Microsoft AI completed pre-training of MAI-Base-2, a mixture-of-experts foundation model with 80 billion active and 2.8 trillion total parameters, and says the recipe is 4.2–12.3x more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base. The run itself is the number worth keeping: 16,000 Nvidia GB200 GPUs for 38 days across August and September, at 21% average model

Amazon's new Bee devices listen all day — for $11.99 a month

Amazon's new Bee devices listen all day — for $11.99 a month

Amazon turned its always-listening Bee line into a two-device family on Thursday, and NVIDIA opened a 2.8-trillion-token Indic corpus recipe. The connecting thread: both are bets that the next AI money is in memory — yours, or your model's. Amazon put two new always-listening devices on sale, and both now lead to a subscription. Bee Frontier, the next-generation wristband that also clips to clothing, ships now at $74.99 in Onyx, Lavender and Sage — up from the $50 predecessor — with seven-day