NVIDIA says Vera Rubin NVL72 has made its MLPerf Inference v6.1 preview debut with up to 3.7x the…
NVIDIA says Vera Rubin NVL72 has made its MLPerf Inference v6.1 preview debut with up to 3.7x the throughput of GB300 NVL72 on Qwen3-VL and up to 2.5x on DeepSeek-R1. Its four-rack GB300 submission also reached 99% scaling efficiency. These are vendor-submitted benchmark…
Deepseek v4.1 flash drawing pixel art.
Not bad though, water is drawn beautifully, but overall couldn't compare to astra/fable yet
I got early access to typesafeai's Jev and have been playing around it.
I think it's safe to say that it won't and cannot replace Astra/Fable/Deepseek, but what it can do is limitless too! If used correctly, this is groundbreaking for bulk heavy, decision making tasks for
Does gemini 4 even exist or are we hallucinating
I asked 4 models across 4 different harnesses to recreate themselves as a WALL-E character 4 models, the exact same prompt, and such different results: - muse spark: built such a cute, polished robot - deepseek: way more
You have reached the end of the archive
All of deepseek