BREAKING: Qatari mediators have reportedly arrived in Tehran for urgent talks to prevent a major regional escalation, conveying Donald Trump’s demands to Iranian officials in a last-ditch effort to avert renewed U.S.-Iran hostilities. Iran is very likely to reject them, having repeatedly refused U.S. conditions and insisting instead on its own terms—including an end to the naval blockade, the release of frozen assets, and a regional ceasefire. The urgent visit comes as Donald Trump warns that the United States could strike Iran as soon as tomorrow, while the USS George Washington carrier strike group continues rearming in the Arabian Sea with bunker-buster bombs and other munitions.

UBS: Server DDR5 and server SSD demand forecasts for 2027 keep being revised higher. We forecast server DDR bit demand up 81% YoY in 2027, server SSD +50% (unconstrained demand x2) $SNDK $MU $SKHY $DRAM $EWY

Reuters: ByteDance, Alibaba and Tencent have ordered more than 3 million of Nvidia’s new RTX PRO 5500 graphics cards with a collective sticker price of over $35 billion, two people familiar with the matter said.

BREAKING: $INTC Intel & $AMD are collaborating to simplify AI development on x86 CPUs via PyTorch. At a recent Open Compute Project panel, Intel confirmed they’re merging oneDNN and AMD’s ZenDNN under a single API so there’s only one upstream path into PyTorch. They’re also jointly advancing the ACE matrix extensions for consistent AI acceleration across both vendors’ CPUs. This reduces fragmentation for developers and improves CPU options for edge and latency-sensitive workloads. Hardware support for ACE is expected later this decade.

oMLX 0.7.1 is out! This is the official 0.7.1 release. It adds GGUF models through Splash, much faster DeepSeek V4.1, 4-bit / 8-bit KV cache formats, and distributed inference for GLM-5.3 and DeepSeek V4.1. What's new - GGUF models through Splash. The downloader filters repos by MLX or GGUF and lets you pick the files, and Qwen3.8-27B and Qwen3.6-35B-A3B GGUFs run on Splash. - Affine4 and Affine8 KV cache. Full-attention KV is 74% smaller with Affine4 at 32K context. - A fix for GPU hangs when several models run at once. - Distributed inference for more models, GLM-5.3 and DeepSeek V4.1 now run across Macs with both pipeline (PP) and tensor (TP) parallelism. - Claude Desktop support. One toggle maps claude-opus, claude-sonnet and claude-haiku to your local models. - Realtime mic transcription with every STT model. Speedups in this release - DeepSeek V4.1 Flash decode on M3 Ultra: 20.9 -> 34.3 tok/s (+64%) for one request, 19.4 -> 91.8 tok/s (+373%) with 8 concurrent requests. - Qwen3.8-Flash-Next with Lightning MTP on M5 Max: decode holds 87-98 tok/s instead of swinging between 64 and 93. - Qwen3.8-Flash-Next with expert offload on M1 Max: +5-8% decode. Coming from 0.7.0? You also get Muse Spark Open, the Recommended models tab, DFlash in the batched engines, decision models, batched prefill and the improved oQe quantization. Thanks to everyone who sent PRs and tested the dev and rc builds!

The more I dig in the various inference engines out there, the more I see the shortcuts they take for speed. Don't assume that model layers are mapped 1-1 into your engine, some are quantizing before running, 8bit and even 4bit. I think I should post more on this, not to finger point anyone, but just to clarify that: more often than not, you bargain precision for speed. Kernel optimizations are great, but there is a physical limit (compute/memory bandwidth). There is no free lunch 🤷🏻‍♂️

Damned if you do, damned if you don't. The reality is for Huggingface to have larger ambitions they need funding, otherwise some companies can threaten them, or say hurt them. Nvidia has done more for Open Source than most companies, and for open weights more than you know.

🇫🇷🇰🇷 Absolutely beautiful. It is reported that the South Korean Air Force's F-35As and the French Air and Space Force's Rafales conducted joint operations in South Korea for the first time in three years. South Korea deployed F-35A, F-15K, KF-16, and FA-50 along with a KC-330 tanker, while France deployed Rafale fighters and an A-330 tanker.