Mirendil Releases Fern 0.1 AI With Massive Context
The new model features a 10 million token context window designed to assist researchers with complex experiments.
Updated on Oct. 26, 2026 in Artificial Intelligence
Live Poll
Do you trust that massive AI context windows are currently reliable enough for complex, high-stakes tasks?
Mirendil has launched Fern 0.1, an artificial intelligence model capable of processing a 10 million token context window. The company is currently accepting requests for early access to the new tool.
Why it matters
The model was developed to help research teams design experiments, fine-tune training runs, and evaluate model performance at scale. Its large context window aims to address limitations in data handling for complex analytical tasks.
Fern 0.1 achieved a 100% success rate on single-needle retrieval tests and 90% on multi-key retrieval tasks at its full 10 million token capacity. The model is a post-trained iteration built upon the Mirendil-Base foundation.
The players
Mirendil
An artificial intelligence company that achieved a $1 billion valuation shortly after emerging from stealth mode.
The details
Mirendil-Base, the foundation model used for Fern 0.1, was built by a single researcher in 3.5 months with training compute costs under $10 million. Access is initially prioritized for enterprise clients and research laboratories.
Timeline
June 2026: Mirendil launched out of stealth with a $200 million seed round.
October 26, 2026: Mirendil officially released Fern 0.1 and opened access requests.
The Tech Race
This development follows a pattern set by the 10 million token context window by significantly extending the memory capacity available for high-level research applications. It positions Mirendil as a competitor in the race to build efficient models with massive data processing capabilities.
Research labs and enterprise users will gain access to the model, potentially accelerating their internal experimental design and model evaluation workflows. Future general availability will depend on the roll-out schedule determined by the company.
The takeaway
The rapid development of the Mirendil-Base model highlights how concentrated research efforts can yield high-capacity tools with relatively low compute expenditure. Researchers should monitor the rollout to see how the model performs in real-world multi-hop reasoning scenarios.
Further reading
Learn more about the latest breakthroughs in Artificial Intelligence.
Source note: This article includes information reported by Startup Fortune.
Live Poll
Do you trust that massive AI context windows are currently reliable enough for complex, high-stakes tasks?
