Compact open models are being fine-tuned for concrete local tasks such as 3D asset generation, vision-language on smartphones, and specialized code output. These releases show a path to useful on-device AI without large datacenter GPUs.
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...
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...
The final release of CCORE GLB-3D-Coder 4B signals a concrete path for generating 3D assets locally: a 4B LLM fine-tuned to produce Three.js code exportable as GLB. GGUF weights and the expected dataset open self-hosted scenarios, with data sovereign...
The final version of CCORE GLB-3D-Coder 4B, a full fine-tune of Qwen3.5-4B, is now available with Safetensors weights and GGUF files. It generates Three.js scenes exportable as GLB for Godot, Blender, and Unity. A concrete signal for self-hosted 3D a...
A Gemma 4 fine-tune directly generates x86_64 ELF files from prompts like 'print boop'. Version 1.1 shows progress on multi-turn and novel strings, but the collapse from Q6_K to Q4_K_M signals a structural limit for local inference. The demo runs in ...
The team behind Carbon-X has released data, code, and the recipe used to turn Qwen3-14B into an open DNA-focused model. The package includes 300B+ midtraining tokens, about 200K SFT conversations, and 98K verifiable RL prompts. For teams handling gen...
