Mirendil-Base trained by one researcher and AI agents
Mirendil says one researcher used AI agents to pretrain the 100B–500B-parameter, code-focused Mirendil-Base from scratch in 3.5 months for less than $10 million, including agent token usage, data processing, and research experiments. The company says agents wrote more than 99% of the code, while the researcher set direction, evaluated results, and intervened at key points; training used public data without purchased datasets, distillation, or synthetic data. Mirendil’s tests reported that Mirendil-Base ranked second behind Kimi K3 on cross-entropy across seven held-out code repositories, ahead of Nemotron-550B-Base, GLM-5.3, and DeepSeek-V4-Pro-Base.
The report presents a concrete example of AI agents handling most implementation work in large-model pretraining, although the claims are company-supplied and have not been independently reproduced.
训练一个5000亿参数的大模型，需要多少人？
AI行业的经验是，至少几十人，大厂动辄投入数百人。
但硅谷一家叫Mirendil的初创公司，把这个数字压缩到了极致——只要一个人。
就在刚刚，Mirendil在官网发了一篇文章，标题非常耸人听闻：《无需预训练团队的大规模预训练》。
文章标题下面列着三个数字：
1名研究员、三个半月、成本不到1000万美元。
#Mirendil#Mirendil-Base#Mirendil-Base-Large#Mirendil-Base-XL#Kimi K3#Nemotron-550B-Base#GLM-5.3#DeepSeek-V4-Pro-Base#PyTorch#JAX