DeepSeek-V4.1-Flash-1017 also tops Opus 5.0 on AutomationBench, 60.5% to 50.3%, and on Apex Bench, 72.1% to 65.8%.

Muse Spark Open and DeepSeek V4.1 Flash both score 39 on Artificial Analysis, yet one costs three times more to run the same ...

Researchers in India have built a unified deep learning framework that classifies brain tumors in MRI scans with 98.52 ...

Teaching a robot to use two arms at once is one of the hardest problems in robotics. Unlike a single arm, bimanual ...

DeepSeek updated V4.1 Flash on October 17. We look at 11 official benchmarks against the previous version and four rivals, ...

DeepSeek V4.1 Flash is currently listed at about 1,500 times cheaper than DeepSeek’s usual off-peak input rate on some third-party hosts. On OpenRouter, ...

Researchers developed an end-to-end inverse design method that uses LSTM prediction, a conditional generative adversarial ...

DeepSeek V4.1-Flash cuts GPU memory for AI agent sessions by 75%, fitting four times as many concurrent sessions on the same accelerator. The Chinese lab's new Causal Encoder-Decoder architecture ...

BEVFormer-Tiny runs its neural-network stages in one Chimera GPNPU kernel. Six camera inputs, 16.1 million cycles per frame ...

DeepSeek, a leading AI enterprise, has announced that its annualized revenue has surpassed 1 billion yuan, marking a significant milestone in its commercial operation. Riding on this strong revenue ...

The 4-node NVIDIA DGX Spark rig pools 512 GB of memory, which is more than enough for the 476 GB footprint of DeepSeek ...

Microsoft's experimental Windows ML update adds a path for running GGUF models. Official documentation and packages show no NPU support, limited generation controls, different distribution ...