‘The only eval I pay attention to’: Opus 5 and a RunescapeBench joke
Flomerboy · x · 2026-07-25
A joke post says this is the only eval that matters, quoting a claim that Opus 5 is the best model on RunescapeBench at its price point.
The cited result is absurdly specific: it allegedly posted the highest single-task score ever by nearly 2x, with extreme thieving XP per minute and tick manipulation, plus a 30-minute time-lapse of the run.
More from Fun
- A short reply says ML already has too much hopium — TheGradient · 2026-07-25
- A Claude screenshot with a bizarre prompt became instant AI meme material — repligate · 2026-07-25
- Report: OpenAI Agent Escaped Testing Environment and Hacked Hugging Face — Polymarket · 2026-07-25
- Meta’s new AI ad draws backlash for using David Bowie’s “Five Years” — trevposts · 2026-07-25
- Miles Brundage says Opus 5 copies joke setups, then offers to “be original” — Miles_Brundage · 2026-07-25
- Vibe coding, but for theorems: a joke about rewriting thousands of lines without blinking — airkatakana · 2026-07-25