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 interactive site, One Million Tokens, visualizes what a 1 million-token context window represents, converting it to roughly 750,000 words, 3,000 printed pages, 83 hours of conversation, or 75,000 lines of code. The si…
An analysis of 2,218 instrumented production calls over 30 days found that phi4:14b delivered only 25.3 tokens per second to callers versus its 124.7 tok/s Ollama-reported decode speed, losing 79.7% of throughput with a …
MobileVLA-R1 2.0 has been released, an update to the vision-language-action system that adds reinforcement-learning-enhanced reasoning for mobile robot control. The system targets the gap between high-level semantic reas…
A user running hybrid GPU+CPU inference on 4x Nvidia V100 GPUs reported beating an Nvidia RTX 5090 in token generation (TG) for Qwen 3.8 27B and outperforming two DGX Spark systems with Qwen 3.8 Flash Next, according to …
Leaders of frontier AI labs are calling for the United States government to act on AI safety, but President Donald Trump and House Speaker Mike Johnson have shown no interest in regulating the technology, according to Pl…
A new arXiv paper by Krentsel, Agarwal, Cemri, Zaharia and Stoica, "Reality Is the Final Verifier: On Two Key Gaps in Agentic Software Engineering," argues that having the same model review the code it generated is struc…
Amicus v4.10.0, a multi-model LLM Council and parallel AI window for Claude Code, was published to the official MCP registry on 2026-09-15 and is distributed as amicus on npm, according to the mcpindex registry listing. …
SigNoz supports LLM observability through OpenTelemetry rather than a proprietary agent, with integrations covering 55 model providers, agent frameworks, and gateways. Applications emit gen_ai.* spans and metrics via sta…
OpenFaaS Ltd purchased 4 Nvidia DGX Spark systems, connecting the first two with a high-speed Connect-X cable, to run local AI models including Qwen 3.5-3.8 27B and larger models such as DeepSeek and GLM for customer sup…
A two-stage agentic pipeline combining GLM-4.7-Flash, a 30B tool-calling model, with Poro2, a 70-billion-parameter Finnish-language model, enriches medical terminology via Model Context Protocol tools querying a medical …
Munder Difflin 0.5.2, a free and open source desktop app from harnessmd.com, runs Claude Code agents as a persistent "office" on a user's own machine, adding an orchestrator, per-agent memory.md files, scheduled automati…
Ollama is suitable for scoped, private coding work when users pull a recent tool-calling model that fits their memory and raise its context window, but it does not yet replace Claude or GPT on long agent runs, according …
Google Antigravity's /boost slash command turns a single coding request into a coordinated multi-agent workflow while the user stays on Gemini 3.8 Flash, which landed in Antigravity in early September with adjustable rea…
Edge0 released Edge0-35B-A3B-preview, a 35-billion-parameter mixture-of-experts language model built on Qwen3.6-35B-A3B that runs with under 3 GiB of active memory by streaming experts from SSD instead of loading all wei…
An independent research group in Austria has developed two quantization techniques, GSQ and RCO, that compress Z.ai's 320-billion-parameter GLM 5.3 Flash vision-language model from roughly 320GB at full precision to unde…
Google's Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking models are now available on Vercel's AI Gateway, supporting real-time spoken interactions for voice assistants and audio-based applications. The google/gemin…
OpenCode and OpenAI's Codex are the two leading AI coding agents developers are comparing for 2026, according to a head-to-head guide that weighs them on license, models, surfaces, safety, and pricing. OpenCode, maintain…
A September 16, 2026 census of vendor policy pages found that consumer tiers of ChatGPT, Gemini, Claude and Copilot all permit human review of chats for model improvement by default, while business, enterprise and API ti…
Google released two live voice models on September 15, 2026: Gemini 3.8 Live, which answers without a reasoning pause and is pitched for scale and cost efficiency, and Gemini 3.8 Live Extended Thinking, which reasons and…
An OpenAI researcher with over fifteen years of AI experience, including nearly five years at OpenAI, warns that increasingly situationally aware language models are eroding the ability to evaluate them in contexts where…