Seraphina Nix (former)

Seraphina Nix

Former Technical Staff

Seraphina Nix worked to design and implement benchmarks for cutting edge models at METR. Previously, she researched adversarial training at Redwood Research and type theory at the Machine Intelligence Research Institute. Outside of work, Seraphina enjoys social dance, rock climbing and learning mathematics.

By Seraphina Nix

Early work on monitorability evaluations

January 22, 2026

We show preliminary results on a prototype evaluation that tests monitors' ability to catch AI agents doing side tasks, and AI agents' ability to bypass this monitoring.

Summary of our gpt-oss methodology review

October 23, 2025

Details on external recommendations from METR for gpt-oss Preparedness experiments and follow-up from OpenAI.

HCAST: Human-Calibrated Autonomy Software Tasks

March 17, 2025

Sharing details about HCAST (Human-Calibrated Autonomy Software Tasks), a benchmark we’ve been developing to measure the abilities of frontier AI systems to complete diverse software tasks autonomously.

Evaluating frontier AI R&D capabilities of language model agents against human experts

November 22, 2024

We’re releasing RE-Bench, a new benchmark for measuring the performance of humans and frontier model agents on ML research engineering tasks. We also share data from 71 human expert attempts and results for Anthropic’s Claude 3.5 Sonnet and OpenAI’s o1-preview.