| Moondream |
https://www.reddit.com/r/LocalLLaMA/comments/1uunqcz/moondream319ba2b/ |
tool |
new |
- |
A small vision-language model. |
| OvisOCR2 |
https://www.reddit.com/r/LocalLLaMA/comments/1uv88co/ovisocr2_a_promising_08b_local_document_parser/ |
tool |
new |
- |
A local document parser model. |
| Mellum2 |
https://www.reddit.com/r/LocalLLaMA/comments/1uv4y2n/mellum2_with_mtp/ |
tool |
new |
- |
An AI model featuring Multi-Token Prediction (MTP). |
| Colibri |
https://www.reddit.com/r/LocalLLaMA/comments/1uv8orn/colibri_streaming_for_hy3_run_hy3_on_10gb_vram/ |
tool |
new |
- |
Streaming implementation for running Hy3 models on lower VRAM. |
| MCP |
https://www.reddit.com/r/LocalLLaMA/comments/1uvaqxp/mcp_is_bad/ |
framework |
new |
- |
Model Context Protocol for connecting AI models to data sources. |
| Flint |
https://www.reddit.com/r/LocalLLaMA/comments/1uv9o2u/studymodels_flint_compressing_reasoning_without/ |
tool |
new |
- |
A compressed reasoning model from StudyModels. |
| FlashMSA |
https://www.reddit.com/r/LocalLLaMA/comments/1uv1f1q/flashmsa_accelerating_milliontoken_training_with/ |
framework |
new |
- |
Acceleration framework for million-token training. |
| Wandancer |
https://www.reddit.com/r/LocalLLaMA/comments/1uvdaq7/wandancer_a_hierarchical_framework_for/ |
framework |
new |
- |
A hierarchical framework for AI agents. |