An experiment by Prediction Log found that a structured evidence-scoring prompt improved Claude's stock-prediction accuracy but hurt Gemma's, leading to the procedure's adoption for Claude and rejecti…
A developer is seeking open-source replacements for Llama 3.3 70B Versatile after Groq shut down the model, focusing on smaller parameter sizes with comparable performance for RAG, agentic workflows, …
ChatOSS, a native desktop workspace for agentic coding built on Ollama, launched as an open-source alternative to Codex, offering AI agents that work inside repositories with an integrated Kanban boar…
Lifehacker reports that running a local large language model (LLM) on a smartphone is now practical for most recent devices, requiring at least 6GB of RAM and using apps like Atomic Chat or PocketPal …
Matteo Grella released Fucina v0.1.0, a CPU-first tensor/autograd library written in pure Zig 0.16, featuring compile-time axis names in tensor types that make misaligned contractions a compile error.…
A developer detailed a sequential model-loading approach to run Whisper, bge-m3, and Gemma on a single local workstation with limited VRAM, avoiding out-of-memory errors by loading and unloading model…
Unsloth, the open-source fine-tuning library, released Unsloth Desktop, a free beta app for macOS, Windows, and Linux that runs and trains LLMs, diffusion models, and audio models locally, directly co…
OpenAI's gpt-oss-120b model, with open weights, requires 72 KiB of KV cache per token in 16-bit precision, calculated from its config.json with 36 layers, 8 key-value heads, and a head dimension of 64…
Google DeepMind's DiffusionGemma, a diffusion-based text generation model, remains highly monitorable despite its latent reasoning capabilities, according to a new analysis that strengthens prior find…
A team at Cua has developed a process-scoped Metal capability shim that unlocks 11-16x faster LLM inference in macOS virtual machines on Apple Silicon, bringing VM performance to 94-99% of bare-metal …
DeepGrove's Maple-Preview 20B-A1B, an open-source ternary-weight reasoning model with a 5.31 GB checkpoint, runs locally on a base Mac mini M4 at 218 tokens per second, according to the company. The m…
Meta released Muse Glimmer, a 30 billion parameter open-weight language model under Apache 2.0, designed for agentic tasks and positioned as a competitor to Qwen 3.6 27B. The model is available as an …
Researchers from Malta Lab introduced LaFaCt, a misinformation detection framework that uses activation engineering to identify false claims by projecting last-token activations onto a 'misinformation…
A Vetted Consumer analysis of 27 recent AI models found that 13 ship under Apache 2.0 and 7 under MIT, while the remaining 7 use five custom licenses with varying restrictions, including revenue thres…
Google Cloud announced the general availability of Autonomous Embedding Generation and AI.SEARCH in BigQuery, along with the public preview of Hybrid Search, to simplify unstructured data analysis. AI…
Knowledge distillation, a method where a smaller 'student' model is trained to mimic a larger 'teacher' model, can produce a student that matches or beats the teacher on specific tasks, according to a…
Google's 2025 open source report highlights 20,000 non-Google contributors, 400 million Gemma downloads, and $2 million in project funding, but the real story is a shared behavioral pattern with Anthr…
DiffusionGemma, a text diffusion model, achieves a 2.2× step-speedup and 31-point GPU compute efficiency gain over autoregressive decoding, according to a performance review by an ML engineer. The com…
Liquid AI released LFM2.5-2.6B, a 2.6-billion-parameter on-device agentic model that outperforms models up to 4x larger on tool use and instruction following, achieving 220 tokens per second on an App…
Researchers introduced PRISMS, a closed-loop framework that detects and steers tool-use failures in agentic LLMs using sparse MLP neuron readouts, achieving ROC-AUC 0.90-1.00 for over-calling and miss…