Model Releases
Who need words any more Words are just human kludge AI don’t need no words
Who need words any more Words are just human kludge AI don’t need no words We release Needle 3: A Sliceable 8-29MB automation foundation model that can match DeepSeek V4 Flash. One set of weights, eve
Who need words any more Words are just human kludge AI don’t need no words We release Needle 3: A Sliceable 8-29MB automation foundation model that can match DeepSeek V4 Flash. One set of weights, every depth from 2 to 20 layers a model of its own, 25-121M parameters at CQ2-bit, built on our Simple Attention Networks and running locally at up to 4k toke…
Related
- Cactus Needle 3: A Sliceable 8-29MB Automation Foundation Model That Matches DeepSeek v4 Flash
- DeepSeek-V4 just dropped on Hugging Face https://huggingface.co/collections/deepseek-ai/deepseek-v4
- Congrats to @deepseek_ai team! Doing the numbers I would estimate: Pro < 14m for the final training run Flash < 4m Ratio of active params …
Source: Emad Mostaque (X) | 2026-09-17