Small fine-tuned and vision-language models are delivering verifiable search, concise answers, 3D assets and on-device vision with low memory overhead. The emphasis is shifting to specialized, predictable and resource-efficient local capabilities.
An early Qwen3-1.7B checkpoint, quantized to Q8 GGUF at 1.83 GB, generates 3D assets from prompts and exports them as GLB for Godot. Aimed at CPU or limited VRAM, it shows how a specialized LLM can bring 3D content generation into a local pipeline, w...
A discrete flow matching framework generates 64³ voxel structures from a single image, passing through intermediate resolutions. JoVox keeps generation lightweight, a relevant feature for those working on local hardware.
A 1.1 GB LLM derived from LFM3-1.7B performs a single search and returns an answer, citation and URL. In tests the base model fails to respond in 62% of live cases, while Obsessive always responds and reduces fabrication. AI-Radar analyzes the trade-...
A fine-tuned LFM3-1.7B acts as a web-search subagent that returns a value, quote, and URL, or "Not found". Trained with LoRA on an RTX 4060 laptop, it cuts invented answers in fixed benchmarks but trades raw live accuracy for reliable replies. Its lo...
Cactus Compute has released Mirage, a 22 MB visual-language model with 51.3 million parameters aimed at phones, wearables, and robots. On a Samsung Galaxy S26 it runs in 74 ms on the NPU and 107 ms on the CPU, covering scene description, object detec...
A lightweight project shows how to cut verbosity in a local LLM. Oc-e-Non-v1, derived from Qwen3-4B-Instruct-2507, reduces average response length by 50% and bakes concise behavior into the model rather than relying only on prompting. That matters fo...
The author of CCORE GLB-3D-Coder has released the dataset used to train the 4-billion-parameter model: 3,918 validated and rendered examples across 40 domains, with reasoning traces, HTML/JavaScript responses, and an MIT license. A base for fine-tuni...
