# Former Anthropic Researchers' Startup Mirendil Releases Fern 0.1, a Coding Model With a Claimed 10-Million-Token Context

On the RULER long-context benchmark at 10 million tokens, Mirendil reports 100% on single-needle retrieval but 45% on multi-hop question answering.

- Published: 2026-10-31T06:16:57.637Z
- Canonical: https://polylog.news/ai/2026-10-31/former-anthropic-researchers-startup-mirendil-releases-fern
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Polylog editors](https://polylog.news), [Behnam Neyshabur on X](https://x.com/bneyshabur/status/2114743262565339148), [Robert Lange on X](https://x.com/RobertTLange/status/2114750025767219693), [MarkTechPost](https://www.marktechpost.com/2026/10/27/mirendil-fern-10m-token-context-model/), [explainx.ai](https://explainx.ai/blog/mirendil-fern-0-1-10m-token-context-coding-agent-model-early-access-claimed-vs-verified-2026)

Mirendil, a startup founded by former Anthropic researchers Behnam Neyshabur and Harsh Mehta, has announced its first model, Fern 0.1 (AI ML Big Data). Neyshabur describes it as pretrained from scratch, with up to 10 million tokens of conte…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-10-31/former-anthropic-researchers-startup-mirendil-releases-fern (subscription information: https://polylog.news/pricing).