Mirendil releases its first model, Fern 0.1, with up to 10 million tokens of contextMachine translation
Mirendil 发布首款模型 Fern 0.1，支持最长 1000 万 token 上下文
Automatically verified and published · Generated and evidence-checked automatically; not reviewed by a human.
Fern 0.1 targets codebases and AI research tasks and can handle about 23 MB of code at once. It is available only through hosted early access, with no model weights released. Mirendil’s reported tests show strong long-context retrieval, but a score of just 6.1% on Terminal Bench 4.0.
This source permits summary display only.
Read at the original source