---
title: "GPT-6 Astra scores 62.7% on ARC-AGI-3 with the standard harness and 99.9% with a new provider adapter harness; Claude Opus 5 scored 30.2%, and GPT-5.6 Sol 7.8% (Greg Kamradt/ARC Prize)"
type: signal
url: https://frontiernewswire.com/s/gpt-6-astra-scores-62-7-on-arc-agi-3-with-the-standard-harness-and-99-ac55c2de
published: 2026-09-03T20:05:01.000Z
byline: "Source: TECHMEME"
---

# GPT-6 Astra scores 62.7% on ARC-AGI-3 with the standard harness and 99.9% with a new provider adapter harness; Claude Opus 5 scored 30.2%, and GPT-5.6 Sol 7.8% (Greg Kamradt/ARC Prize)

Source: TECHMEME · Published 2026-09-03 · Canonical: https://frontiernewswire.com/s/gpt-6-astra-scores-62-7-on-arc-agi-3-with-the-standard-harness-and-99-ac55c2de

Greg Kamradt / ARC Prize : GPT-6 Astra scores 62.7% on ARC-AGI-3 with the standard harness and 99.9% with a new provider adapter harness; Claude Opus 5 scored 30.2%, and GPT-5.6 Sol 7.8% — Summary — GPT-6 Astra scores 62.7% for $26K on ARC-AGI-3 Semi-Private with our Standard harness, and 99.9% for $19K with a Provider Adapter harness.

In this story: [GPT-5.6 Sol](https://frontiernewswire.com/models/gpt-5-6-sol), [Claude Opus 5](https://frontiernewswire.com/models/claude-opus-5)

## Sources

- [Read the original at TECHMEME](https://www.techmeme.com/260903/p40#a260903p40)

## Related on Frontier

- [Researchers say LLMs can be good at natively reading and writing neural network weights; GPT-6 Astra edited Qwen3.5-0.8B's weights to make it better at math (Rikiya Takehi)](https://frontiernewswire.com/s/researchers-say-llms-can-be-good-at-natively-reading-and-writing-e4f14b26)

---

[More Latest Signal](https://frontiernewswire.com/) · [Signals](https://frontiernewswire.com/verified-reporting)