Latest

Context language models are language models that manage their own context natively. They treat working memory as a file they're free to edit, instead of a transcript that only ever grows. September 30, 2026. That's the day Facebook Research's Context Language Models showed up

Deadline first, motivation second. That is the usual order for anyone who keeps meaning to get hands-on with open-weight models and never does. The interest exists, the date does not. October 2026 supplies the date. More than 300 community events, in person and online, all aimed at one skill. Grounding

MF-1 posted 82.8, and multimodal flow modeling just moved out of my "huh, interesting" folder and into my architecture notes. One model, trained on 150B pretraining tokens, generating text and images under a single training objective. If your current stack is a language model wired to an

The new benchmark capped frontier AI agents at 68% accuracy. And that number should change how you pick an agent framework. The benchmark, posted to arXiv on September 29, 2026 as paper 2609.38137v1 by Quang Hieu Pham, Thuy Duong Nguyen, Jocelyn Qiaochu Chen. And Xi Ye, runs a stress

LongHarness Bench landed on arXiv September 29, 2026, and the best score in the whole paper is 68%. That's the macro-average accuracy of the strongest model-runtime pairing across four evaluation suites. And it's the number that should recalibrate how you buy agent tooling this year. If

KV-streams landed on arXiv September 29, 2026, with a blunt claim: training agents that run long stops choking on GPU memory once you stop throwing the cache away. The paper, "KV-streams for Efficient Compaction in Agentic Reinforcement Learning" (arXiv 2609.35750), reports a 2.6 to 5x wall-clock

Phone version of Meta Muse already live. Glasses version queued for "the coming months." Between those two states sits the only window a small store will get to prepare. And most of the launch coverage is spending that window on the wrong half of the story. What Meta