SRE-Bench (four attempts)

Updated Sep 5, 2026·2 views

Problem Count: —
Institution: OpenAI
Category: Coding and Software Engineering
Metrics: Score (%)
Language: English
Difficulty: —

Overview

Success within four attempts; not single-attempt success.

Related resources

Official Website

Latest SRE-Bench (four attempts) model rankings and full benchmark leaderboard

Browse the latest scores, model modes, release dates, and parameter sizes for SRE-Bench (four attempts).

Source: DataLearnerAI

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Model Mode Legend

License:

Origin:

Model release cutoff:

Rank	Model				License
	GPT-6 Astra unknown	99.20	2026-09-03	Unknown	Closed
	GPT-5.6 Sol unknown	68.70	2026-06-26	Unknown	Closed

Latest SRE-Bench (four attempts) model rankings and full benchmark leaderboard

SRE-Bench (four attempts) Rank