Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
A three-day comparison of AI web scraping tools found Firecrawl, at roughly $20 per month for its starter tier, outperforms Browse AI (~$40/mo) and a custom Playwright plus Claude 3.5 Sonnet loop for developers, because …
Google launched Gemini 3.8, its latest conversational AI model, which the company says enhances user interaction by thinking and handling tasks without disrupting the flow. The release is part of Google's effort to stren…
Google released Gemini 3.8 Live Extended Thinking, an audio-to-audio model that performs background reasoning and asynchronous tool calls while streaming continuous audio, with a 131,072-token input limit and 65,536-toke…
AirmailAI launched as an open-source, bring-your-own-key (BYOK) LLM chat app that runs as a browser extension backend on desktop Chromium-based browsers such as Chrome, Edge, and Brave. The app supports OpenAI, Anthropic…
GoogleDeepMind unveiled Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, two audio models built to enable natural voice applications that let users perform complex tasks while maintaining an uninterrupted conversat…
A developer built Cituna, an answer engine optimization tool exposed as a Model Context Protocol (MCP) server that queries buyer prompts across seven AI answer engines—ChatGPT, Perplexity, Gemini, Claude, Grok, Google AI…
Google introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, two models built for near real-time reasoning and production-ready voice agents. Gemini 3.8 Live Extended Thinking took the #1 overall spot on Arti…
NVIDIA's Nemotron 3.5 Lightning 30B-A3B model demonstrates how a Mixture-of-Experts (MoE) architecture can activate only 3B of its 30B parameters per token by routing each token through a selected subset of expert feed-f…
Salesforce and NVIDIA unveiled Koa, a CRM domain-specific reasoning model for Agentforce, at Dreamforce Conference 2026, with Salesforce president and chief platform and engineering officer Rohan Kumar saying it is built…
The Office of Personnel Management will continue deploying ChatGPT and five other large language models after the General Services Administration announced a new OneGov partnership with OpenAI offering agencies reduced c…
A developer guide details how to optimize local large language models using Ollama, noting that models such as Qwen 3.5 and Gemma 4 now handle coding, function drafting, and document summarization with surprising accurac…
A developer using the handle robotman105 has released Compute-Ranch, a zero-config personal compute cluster program, on GitHub. The software pools Linux machines into a single job queue with watch-folder automation and c…
A developer published a Bash launcher script, claude.sh, that routes Anthropic's Claude Code CLI between Google Vertex AI and an internal "dogfood" AI gateway. The script selects a route via the first argument, expands s…
Amazon Web Services published a walkthrough for building an AI-powered product tagging system that customizes Qwen3-8B with supervised fine-tuning (SFT) and then optimizes it with reinforcement learning with verifiable r…
JetBrains released the Kotlin Benchmark, an official SWE-bench-style evaluation that grades AI coding agents on 105 real repository-level Kotlin engineering tasks, with Claude Code running Opus 4.7 xhigh leading the firs…
Crusoe announced a multi-year partnership with Perplexity on September 15, 2026, under which Perplexity will train frontier models on Crusoe's dedicated NVIDIA GB300 NVL72-powered clusters and NVIDIA InfiniBand Solutions…
IBM Research introduced consistency guidelines in its altk-evolve toolkit, built on a new Consistency Analyzer, to address a 24.4-point consistency gap in LLM agent reliability. A ReAct agent using GPT-4.1 on AppWorld's …
Glad Labs decided not to switch its self-hosted inference stack from Ollama to vLLM after reviewing its own call logs, which showed only one to three concurrent calls at most against roughly 50 calls a day on an ASUS ROG…
Few-shot prompting with three to five high-quality examples can fix classification errors that zero-shot prompts produce in LLM applications, according to a developer account published on PromptCube. The author reported …
LangChain's LangSmith observability platform can record every step of non-deterministic AI agent workflows by organizing data into OpenTelemetry-like runs and trace trees, according to a technical walkthrough published b…