@Mnilax: China f*cked the world again DeepSeek just dropped a killer library that saves up to 98% on tokens ur GPT Astra and Opu…
Summary
DeepSeek released DeepGEMM, an open-source GPU math engine used to train and run its own models, claiming up to 98% token cost savings by squeezing more tokens per second out of existing GPUs.
View Cached Full Text
Cached at: 10/05/26, 01:21 AM
China f*cked the world again
DeepSeek just dropped a killer library that saves up to 98% on tokens ur GPT Astra and Opus 5.5
it’s DeepGEMM, the engine DeepSeek trains and runs its own models on
faster GPU math -> more tokens per GPU -> cheaper tokens
plug the engine DeepSeek uses into your own models squeeze more tokens per second out of the GPUs you already pay for match or beat expert-tuned GPU libraries on speed
u will pay $1.6 instead of $50
its sitting in one public repo, link below
Similar Articles
@Ryrenz: Shocking news: DeepSeek has open-sourced DeepGEMM, the GPU kernel library they used to train their models, squeezing ev…
DeepSeek 开源了其训练模型所用的 GPU kernel 库 DeepGEMM，将 FP8/FP4/BF16 GEMM、融合 MoE 通信重叠、MQA 打分等核心算子统一到单一 CUDA 代码库中，通过 DeepJIT 运行时编译，官方称性能可达 H800 上 1550 TFLOPS，并已推出昇腾（Ascend）版本。
@cyrilXBT: CHINA JUST BUILT AN AI MODEL THAT IS COMPETING WITH OPENAI AND ANTHROPIC AT A FRACTION OF THE COST. And someone just dr…
DeepSeek, a Chinese AI model built by a quant hedge fund, is reportedly competing with GPT-4 level performance at roughly 5% of the training cost, causing significant market disruption including a $600B drop in NVIDIA's market cap. A free 1 hour 50 minute course has been released teaching users how to leverage DeepSeek V4 locally and via API.
deepseek-ai/DeepGEMM
DeepSeek releases DeepGEMM, a high-performance CUDA kernel library for LLM computation primitives including FP8/FP4/BF16 GEMMs, fused MoE with overlapped communication, and MQA scoring, compiled at runtime via JIT with no installation-time CUDA compilation required. The library achieves up to 1550 TFLOPS on H800 and matches or exceeds expert-tuned libraries across various matrix shapes.
DeepSeek R2 just went open-source and it's matching GPT-4o on 9 of 12 benchmarks — for literally $0 in API costs
DeepSeek R2, a new open-source model, matches GPT-4o on nine of twelve benchmarks while running locally on a single A100 for zero API cost, potentially transforming the economics of AI deployment.
@seclink: Chinese startup DeepSeek announced on Friday that its 75% discount on the DeepSeek-V4-Pro API will become permanent, with prices as low as $0.003625 per million cached input tokens and $0.87 per million output tokens—approximately 34 times cheaper than OpenAI's GPT-5.5. The model has 1.6 trillion...
DeepSeek has made its V4-Pro API price cut of 75% permanent, with per-million cached input tokens at just $0.003625 and output tokens at $0.87, about 34 times cheaper than OpenAI's GPT-5.5. The model has 1.6 trillion parameters but requires only 49 billion active parameters, supports a 1-million-token context, and leads in coding and reasoning benchmark tests.