Risk Evaluation · Benchmark

MirrorCode: Evidence that AI can already do some weeks-long coding tasks

NUWA research note · independent review pending

First published: 2026-04-10Source checked: 2026-10-25

Original source ↗Cite this record ↓Autonomy →

NUWA Research Note

Research question Design and assumptions Source findings NUWA interpretation Limits and contrary evidence Research and decision relevance Institutions and roles Original resources

Model-assisted reading · Selected sections of the original full text or official article; a complete section-by-section audit remains pending.

At a glance

NUWA interpretation: precise behavioral specifications enable long coding tasks that differ from open-ended project work.

Research question

Can agents reimplement whole CLI programs from behavioral specifications?

Design and assumptions

The Epoch article provides execute-only reference access, documentation and test-based grading, while hiding source and web access; paired hidden tests discourage hard-coding.

MirrorCode — Source figure — Source figure · Original source figure. Read alongside the protocol scope and limitations below. Original figure and source ↗

Source findings

Recent models fully reimplemented several tested programs; Pkl remained unsolved in the preliminary set.

Opus 4.6’s progress on the larger gotree target exceeded earlier models.

NUWA interpretation

NUWA interpretation: precise behavioral specifications enable long coding tasks that differ from open-ended project work.

Limits and contrary evidence

The authors identify detailed-spec requirements, possible contamination and a small/narrow target set as key limitations.

Research and decision relevance

Model recommendation: cite Epoch’s methods, rather than treating METR’s linkpost as a full report, and retain preliminary status.

Precise source locations

MirrorCode: Evidence that AI can already do some weeks-long coding tasks · MethodologySelected original resource https://epoch.ai/publications/mirrorcode-preliminary-results/; sha256:69894ca02d92cd62e39467c215b90d749bd087db5b0966435d921e95896166ff; archived source capture; exact capture time not independently established
MirrorCode: Evidence that AI can already do some weeks-long coding tasks · Preliminary resultsSelected original resource https://epoch.ai/publications/mirrorcode-preliminary-results/; sha256:69894ca02d92cd62e39467c215b90d749bd087db5b0966435d921e95896166ff; archived source capture; exact capture time not independently established
MirrorCode: Evidence that AI can already do some weeks-long coding tasks · LimitationsSelected original resource https://epoch.ai/publications/mirrorcode-preliminary-results/; sha256:69894ca02d92cd62e39467c215b90d749bd087db5b0966435d921e95896166ff; archived source capture; exact capture time not independently established

Source checking, extraction, editorial approval and replication are recorded separately. Methodology →

Institutions and roles

Published by METR

Evaluation protocols

MirrorCode →

Cite and reuse

Cite the original publication for its findings. Cite this NUWA record separately for its explanation and retained snapshot.

BibTeX ↓RIS ↓Record + source metadata · JSON ↓Reading coverage · JSON Suggest a correction

Original publication · BibTeX

@misc{mirrorcode,
  title = {MirrorCode: Evidence that AI can already do some weeks-long coding tasks},
  url = {https://metr.org/blog/2026-04-10-mirrorcode-preliminary-results/},
  year = {2026},
  date = {2026-04-10},
  version = {Source capture 2026-10-25},
  note = {Source capture 2026-10-25}
}

NUWA record · BibTeX

@misc{mirrorcode-nuwa,
  title = {MirrorCode: Evidence that AI can already do some weeks-long coding tasks — NUWA research record},
  url = {https://whitzard.tech/en/nuwa/whitzard-index/work/mirrorcode/},
  version = {2026-10-29.1},
  organization = {NUWA Frontier AI Safety Lab},
  howpublished = {Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--mirrorcode.json},
  note = {NUWA research note. Source-reading scope and independent review status are stated on the record. Snapshot 2026-10-29.1. Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--mirrorcode.json.}
}

Evidence snapshot 2026-10-29.1 · Index updates · RSS

References

[1] MirrorCode: Evidence that AI can already do some weeks-long coding tasks. 2026-04-10 ; Official publication masthead and article byline; Methodology; Preliminary results