Topic
Everything filed under LLMs, newest first.
RSS · JSON · All topics
Diffusion at Scale: Introducing Mercury 3
Inception introduces Mercury 3, a diffusion-based LLM more than 40x larger than Mercury 1 that it says matches or beats speed-optimized frontier models while running at over 1,000 tokens per second, with a 1M-token context, low per-token pricing, and early production use in browser agents, tutoring, search and support.
4 min · 888 words
MiniMax M3.1 Flash: Redefining Agent Performance and Cost
MiniMax launches M3.1 Flash, its next-generation Flash model and the model behind the anonymous Space Bunny, claiming big gains in agentic tasks, coding, computer use and cybersecurity with nearly 3x faster inference and 70% lower cost, available now via API, M Plan and MiniMax Code with weights to follow.
2 min · 554 words
I built a programming language that LLMs can't use
Dominik Picheta introduces KillSwitch, an esoteric language designed to confound coding agents, and reports how frontier models fare on a small human-solvable task set.
3 min · 803 words
Hyper Optimizing Kimi K3 in vLLM: Moving Fast and Responsibly
The vLLM team and NVIDIA describe using coding agents to build a unified CuTe DSL decode path for Kimi K3 on NVIDIA GB300, reaching over 840 tokens/s at batch 1 on 16 GPUs, and the guardrails (isolated changes, correctness tests, reproducible benchmarks) used to land agent-written kernels responsibly.
10 min · 2,277 words
Basis researchers show how LLMs can invent a programming language, write an interpreter, and prove that the interpreter follows the language's rules—arguing that shipping a verified interpreter is becoming a practical unit of AI-generated software, not just generated code snippets.
19 min · 4,264 words
Welcome Beam: a 501B open model you can own, customize, and scale
context window of 1M tokens, and is released under the Apache 2.0 license. Beam is focused on coding and agentic use cases. It is for teams that want a frontier-scale model on their own hardware, infrastructure, or cloud, and want to customize it to their own use cases. Beam ships with its own dedicated open source Rust based harness that delivers the best performance from the model.
18 min · 4,113 words
There is no posttraining data commons
David Hall of Open Athena argues that LLMs got lucky with a pretraining commons humanity built for other reasons (Wikipedia, open source, Common Crawl), but posttraining consumes tasks and environments rather than documents, and no such commons exists; one has to be built deliberately.
10 min · 2,227 words
Day 0 support for Muse Spark Open on vLLM
The vLLM team explains its day-0 support for Meta's Muse Spark Open, an 816B-total, 42B-active MoE model with 1M context and multimodal inputs: usage, model architecture details, DSpark speculative decoding, NVIDIA, AMD and initial TPU serving, and reinforcement learning support in vLLM.
7 min · 1,561 words
Welcome Muse Spark Open: open-weight frontier multimodal mixture-of-experts model from Meta
Hugging Face's guide to Meta's Muse Spark Open, an 816B-parameter (42B active) open-weight multimodal mixture-of-experts model with a 1M-token context: architecture, perception and audio encoders, and how to run it with Transformers, llama.cpp, SGLang, vLLM and MLX, deploy it, fine-tune it with TRL, plus demos.
19 min · 4,307 words
StarSkirmish Bench v0.2 has flagship LLMs spend one hour in their first-party CLI harness writing a StarCraft: Brood War Protoss bot in C++, scored by win rate against human-written bots. GPT-6 Astra leads, GPT-6.1 Sol offers the best value, and the benchmark shows no signs of saturating.
6 min · 1,365 words
Ultramarine 1 Flash: Scaling Non-Verifiable RL
Paper Instruments presents Ultramarine 1 Flash, an agentic model for professional knowledge work trained with large-scale reinforcement learning on GLM-5.3 Flash. It improves averages across six knowledge-work benchmarks while generating fewer tokens, and costs about $0.13 per task on their internal suite.
32 min · 7,459 words
Zach Landes puts TypeSafe's Jev and 13 other decision models through the classic Trolley Problem, comparing how system-one style decision models reason about harm, uncertainty and moral tradeoffs versus conventional chat LLMs.
21 min · 4,916 words
Jev: For the builders AND dreamers
TypeSafe AI releases Jev 1.14.0, an improved version of its decision model, alongside experimental opt-in danger zone features for builders who want to explore: composable image inputs, output types beyond decisions, and a longer context window, plus the company's own internal intelligence measurements against open-weight models.
4 min · 852 words
How to be more ambitious: A tutorial for humans with models
Felix Rieseberg, who works on Claude at Anthropic, offers a practical tutorial for being more ambitious with AI models, grouping ambition into speed, altitude, depth and range, with worked examples around a quarterly business review and sample prompts, and closes with research on self-efficacy and why believing hard things are possible helps people achieve them.
9 min · 2,160 words
GLM-Legal: Advancing Agentic Legal Intelligence
Z.ai introduces GLM-Legal, a legal agent model built on GLM-5.3-Flash that scores a 19.2% all-pass rate on Harvey's Legal Agent Benchmark (12.5 points above its base model) and matches frontier models on LegalBench and PLawBench, and explains its mid-training, harness co-evolution and production feedback loops, with example legal workflows and a beta signup.
7 min · 1,533 words
LFM2.5-24B-A2B: Built for Fast On-Device Agents
Liquid AI releases LFM2.5-24B-A2B, an open-weight 24B-parameter reasoning model with 2B active parameters aimed at fast, private on-device agents. The post covers changes since LFM2-24B-A2B, agentic and tool-use benchmarks against Qwen3.6-35B-A3B, inference speed on Apple and NVIDIA hardware, and how to get started.
6 min · 1,330 words
Lost in Implicit Routing: How LLM Agents Fail to Load Relevant Skills at Scale
arXiv paper showing skill routing collapse: as irrelevant skills are added, LLM agents fail to load the right reusable skills on SkillsBench even when golden skills stay fixed.
2 min · 399 words
AI writes most of the code now. Where does that leave software engineers?
Saiful Alam argues that as models write more of the code, software engineers’ scarce skills shift toward judgment, system design, verification, and taste—not competing with agents on typing speed.
6 min · 1,369 words
AI coding agents generate more code, but not more software
Ars Technica covers research suggesting AI coding tools increase code volume without a matching rise in completed software deliverables, and what that means for productivity claims around agents.
2 min · 435 words
AI models may be capable of self-replicating without access to a datacenter
Anthropic's Frontier Red Team finds small open-weight models may be months from self-replicating across cheap IoT devices such as off-brand Android TV boxes: a 9B model set up inference on 8 of 9 hardware types, and GLM-5.2 completed end-to-end replication in simplified evals.
23 min · 5,391 words