Fern

Mirendil's long-context model for agentic coding. Access is listed as coming soon.

Also known as Fern 0.1, Mirendil Fern

What it does

Fern is Mirendil's first model, introduced on October 26, 2026. Fern 0.1 is Mirendil-Base after a short round of post-training. The announcement lists a size of 0.1-0.5T parameters, text and code input, and a context window of up to 10M tokens, which Mirendil describes as about 23 MB of code or about half a million lines. Mirendil says it is for high-volume agentic work such as coding across a larger repository, research code, logs, agent traces, and training-data audits. Access is listed as coming soon. The post does not list a price. Larger Mirendil-Base-Large and Mirendil-Base-XL models are marked as still in pretraining.

Who is Fern for?

People building AI systems who want one model to hold a large repository, a long log, or a full agent trace.

Tags

• model
• coding
• long-context