The State of Context Engineering(dbreunig.com)
220 points by pchristensen | 57 comments
Slides and transcript of Drew Breunig's AGNTCon talk "Context Fails & Fixes," reviewing the history, failure modes, and future of context engineering for LLMs.
#long-context on Hacker News
9 stories tagged long-context, newest first.
Step 5 Preview, a 1M-context MoE from StepFun, shows up on OpenRouter(openrouter.ai)
133 points by AnneWodell | 33 comments
OpenRouter model page for StepFun's Step 5 Preview, a 600B-parameter sparse MoE model with 1M context, multimodal input, and $1/$2.70 per million token pricing.
LensVLM: Compressing long context as images, expanding only relevant pages(huggingface.co)
77 points by victormustar | 7 comments
Hugging Face model card for Apple's LensVLM-9B, a vision language model that reads compressed images of text and expands only relevant pages via learned tools.
GLM 5.2 Is Out(twitter.com)
701 points by aloknnikhil | 413 comments
Announcement that Z.ai released GLM-5.2, a coding-focused model with a 1M-token context window, with an MIT-licensed open release planned for next week.
Cerebras launches Qwen3-235B, achieving 1.5k tokens per second(cerebras.ai)
334 points by mihau | 140 comments
Cerebras press release announcing Qwen3-235B on its Inference Cloud with full 131K context support, billed as the world's fastest frontier AI model.
Smollm3: Smol, multilingual, long-context reasoner LLM(huggingface.co)
356 points by kashifr | 71 comments
Hugging Face blog post announcing SmolLM3, a small multilingual, long-context reasoning language model, published July 8, 2025.
The killer app of Gemini Pro 1.5 is using video as an input(simonwillison.net)
1074 points by simonw | 444 comments
Simon Willison's blog post on Google's Gemini Pro 1.5, arguing its standout feature is video input, demonstrated by uploading a bookshelf video via Google AI Studio.
Our next-generation model: Gemini 1.5(blog.google)
1192 points by todsacerdoti | 553 comments
Google blog post by Sundar Pichai and Demis Hassabis announcing Gemini 1.5, a model with improved performance and long-context understanding across modalities.
From deep to long learning?(hazyresearch.stanford.edu)
451 points by headalgorithm | 98 comments
Stanford Hazy Research blog post by Dan Fu, Michael Poli, and Chris Ré on their lab's work increasing sequence length in deep learning models.