A developer compared Llamafile and vLLM for serving local LLMs on an RTX 4090, finding Llamafile faster for single-user interactive use and simpler to set up, while vLLM excels under concurrent load a…
HuggingFace released Tau, an open-source Python coding agent designed to be easily understood, with its core agent loop fitting in about 200 lines of code. The project, named after the circle constant…
Anthropic's June 2026 Economic Index report, based on millions of Claude conversations and a survey of 9,700 users, reveals that over one-third of workers expect AI to handle most of their job tasks w…
A developer trained a DistilBERT model on the AG News dataset in 10 minutes using a rented GPU and the Crunr CLI, demonstrating a cold-start fine-tuning pipeline. The experiment highlights the importa…
Surge AI released GDP.pdf, a benchmark testing frontier AI models on real-world PDF tasks across ten domains, and found every model scored under 30%. The benchmark highlights critical failures in fina…
A developer built Draco AI, a full-stack, hardware-agnostic LLM system from scratch in pure Python and NumPy, eliminating heavy framework dependencies like PyTorch or HuggingFace. The system is 100% o…
ZML released ZML/LLMD, an inference server written in Zig that runs LLaMa, Gemma, Qwen, and Mistral LLMs on five architectures including NVIDIA CUDA, AMD ROCm, Google TPU, Intel oneAPI, and Apple Meta…
BareMetalRT launches TensorRT-LLM natively on Windows without WSL, enabling heterogeneous tensor parallelism across consumer GPUs over standard networking. Users can run HuggingFace models locally via…
NVIDIA released Nemotron 3 Ultra, a 550-billion-parameter open-weight MoE model with hybrid Mamba-2 and Transformer architecture, under the Linux Foundation's OpenMDW-1.1 license. The model requires s…
Kotro Labs released Kotro Proxy Engine, an open-source Rust sidecar that acts as a local AI governance layer to reduce token usage and prevent runaway costs in agentic coding workflows. The tool featu…
JetBrains released Mellum2, a 12-billion parameter open-source Mixture-of-Experts model that activates only 2.5 billion parameters per inference, making it twice as fast as equivalent models. Designed…
Researchers from General Intuition, Kyutai, and Epic Games released MIRA, a 5-billion-parameter latent diffusion world model that generates real-time 2v2 Rocket League matches at 20 FPS on a single GP…
A developer added opt-in GPU backends (Metal and CUDA) to a pure-C TTS inference engine for Qwen3-TTS, achieving faster-than-real-time performance on Apple Silicon and NVIDIA GPUs. The key optimizatio…
BottleCap released the Thinking CAP model series, a fine-tuned version of Qwen3.6-27B that reduces reasoning tokens by 46% while maintaining benchmark performance. The model aims to cut inference cost…
LangChain released the fourth installment of its learning series, explaining how to build AI workflows using Chains and the LangChain Expression Language (LCEL). The article demonstrates Simple, Seque…
Álvaro of Punky Tiger Labs has developed a persistent AI that maintains memory, identity, and continuity over time, unlike stateless LLMs. The AI powers a 24/7 autonomous radio show, but the team cann…
A group of AI researchers launched FLARE-AI, a crowdsourced website for reporting and tracking AI harms, enabling users to flag issues like malware generation or privacy leaks. The open-source system …
A developer with ADHD built an AI agent called 'hex' to help manage executive dysfunction by integrating with calendar, Todoist, Obsidian, web search, browser, TTS, and Telegram. The agent uses memory…
DeepSeek V4 Flash, a 284-billion-parameter Mixture-of-Experts model, now runs locally on a 128GB MacBook Pro via antirez's experimental llama.cpp fork, achieving ~21 tokens/sec generation on the Metal…
A new audit of over twenty African NLP corpus families reveals widespread license incompatibilities, including CC-BY-SA and CC-BY-NC datasets that cannot be legally combined and NoDerivs clauses that …