Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
An anonymous email attributed to a machine learning engineer at a frontier AI lab claims the lab has hit limits on scaling models for programming tasks and that frontier labs lack a competitive moat, predicting an open-w…
Richard Socher's new venture Recursive raised a $4.65 billion seed round to build a "Eureka Machine" aimed at automating AI research itself, with Socher claiming a Recursive AI research system outperformed humans and the…
Elon Musk said on September 14, 2026 that Grok 4.8, a 2.5T model trained with a new C++ software stack, will finish training that week and start RL. The post from Musk's account drew 2.8 million views. The claim concerns…
A developer documented building a fully offline conversational chatbot named Daisy using Ollama, LangChain, and Streamlit on a CPU-only laptop with 16GB of RAM and no dedicated GPU. The project runs Meta's llama3.2 3B mo…
Cursor outperforms GitHub Copilot for Python projects with deep codebases because it indexes the entire local project, letting it reference installed library versions and types across 20+ modules, while Copilot is limite…
Tencent Yuanbao, Tencent's free consumer AI assistant built on the Hunyuan family, has been added to the Uprouter directory, which rates it as medium risk. The current version serves the Hunyuan Hy4 preview with upgraded…
Tencent AI Studio, the company's free consumer web studio for its Hunyuan (Hy) model family, has been added to the Uprouter directory as an unofficial web-cookie integration using session cookies from aistudio.tencent.ai…
Uprouter added Qwen Web (Free), Alibaba's free consumer chat app for the Qwen family at chat.qwen.ai, to its API directory, listing the cheapest tracked model at $0.030 per 1M input tokens and $0.130 per 1M output tokens…
Uprouter added Muse Spark Web (Meta AI), an unofficial web integration of Meta's free consumer assistant at meta.ai, to its provider directory. The listing states the service runs on Meta's hosted multimodal reasoning mo…
Arena (Free), the free anonymous multi-model chat and blind A/B voting service behind the public LLM, image, and code leaderboards at arena.ai (formerly LMArena/Chatbot Arena), has been added to the Uprouter provider dir…
Uprouter added Hailuo Web (MiniMax), MiniMax's consumer chat product at chat.minimax.io, to its model directory, listing MiniMax-01 at $0.200 per 1M input tokens and $1.10 per 1M output tokens as the cheapest tracked mod…
Uprouter added Gemini Web (Free), an unofficial browser/cookie integration that forwards to Google's hosted Gemini models, to its API directory. The listing rates the provider as medium risk because it relies on web-cook…
Bolt.new co-founder and CEO Eric Simons launched Bolt Forge on September 14th, giving paid individual users up to 50 times their usual AI-building allocation in exchange for opting into a training-data program with Arcee…
Andon Labs released Pion, an agent designed to run any company fully autonomously, opening it to a waitlist after two years of research and real-world deployments of autonomous businesses including vending machines, a st…
A developer argues that benchmark saturation and shifting evaluation methods make it increasingly difficult to judge how much better new AI models like GPT-6 Astra actually are, citing Astra's system card examples of ret…
Apple released Xcode 27, the first version to drop Intel Mac support entirely, requiring Apple Silicon hardware to build for iOS 27 as of today. The release ships with Swift 6.4, native support for Claude, Gemini, and GP…
A developer built both single-agent and multi-agent versions of an LLM-powered customer support system and ran them through an identical eval suite, finding no difference across five properties including safety, intent a…
A developer has released Milldown, a native macOS app that converts PDFs, Office documents, spreadsheets, web pages and other formats into clean Markdown entirely on the local machine for LLM and RAG pipelines. The app w…
Perplexity released Portable Computer for Windows, a local AI agent that runs on NVIDIA GeForce RTX or RTX PRO GPUs with at least 24GB of VRAM and uses a Qwen 3.8 27B model post-trained for the Perplexity Computer enviro…
NVIDIA reported that combining its Transformer Engine library with the JAX Python library raised DeepSeek-V3 MoE training throughput on NVIDIA GB200 GPUs from an unoptimized baseline of 103 TFLOPS/GPU to 1,068 TFLOPS/GPU…