Skip to content
Trending storyDeveloping

Dex Horthy reports GPT-6 Astra leads SlopCodeBench at 16.3% strict pass rate

2 articles1 sourcesince Oct 27Last article

Overview

AISummary of 2 articles

Dex Horthy says GPT-6 Astra scored a 16.3% strict pass rate on SlopCodeBench in October 2026, the highest among the frontier models he ran.

He cites GPT-5.5's 14.8% from the May SlopCodeBench publication as the comparison. Horthy also says Opus 5.5 scored better at medium effort than at xhigh, while Astra and SOL scored better at xhigh.

Horthy says the full 36-problem, 196-checkpoint run will appear on the main SlopCodeBench leaderboard soon. A later post from him announces new SlopCodeBench results in his livestream "AI That Works #79," but that post gives no scores or models, and the full run had not been published at the time of reporting.

Written by AI from the articles below · updated Oct 27, 1:43 PM ET

Check the sources:

Latest developmentOct 27, 1:20 PM ET
Dex Horthy shares new Slopcode Bench results in livestream

Developments

2 developments

  1. Oct 27, 1:20 PM ET · 1 article
    Dex Horthy announces new Slopcode Bench results in AI That Works #79
    Dex Horthy: Dex Horthy shares new Slopcode Bench results in livestream
  2. Oct 27, 12:16 PM ET · 1 article
    Dex Horthy reports SlopCodeBench results with GPT-6 Astra leading at 16.3% strict pass rate
    Dex Horthy: SlopCodeBench shows Astra leads on strict pass rate at 16.3%

Article timeline

The articles in this story. Times are ET.

Oct 27
  1. dexX
    Dex Horthy shares new Slopcode Bench results in livestream

    AIDex Horthy announces new Slopcode Bench results in the livestream "AI That Works #79." The post links to the broadcast and gives no benchmark scores, models, or other figures.

  2. dexX
    SlopCodeBench shows Astra leads on strict pass rate at 16.3%

    AIDex Horthy reports that gpt-6 astra scored a 16.3% strict pass rate on SlopCodeBench, the highest among frontier models tested in October 2026, up from gpt-5.5's 14.8% in May. Opus 5.5 scored better on medium effort than on xhigh, while Astra and SOL performed better on xhigh. The full 36-problem, 196-checkpoint run is expected to appear on the main SlopCodeBench leaderboard soon.

Heat Trend

Current comparable heat
6
Peak comparable heat
10Oct 27, 1:00 PM ET
Change in past 24 hours
–

The trend compares only continuously observed participants, so its scope may be smaller than the current heat total. Hover or tap for hourly heat; use arrow keys with a keyboard.