Frontier Log

Benchmark · Active

FrontierSWE v2

Created by Proximal, 2 Sep 2026 · Higher is better · Website

020406080Oct 202669.9%Fable 5.261.2%Gemini 4 Carbon55.0%Gemini 4 Argon61.9%Sonnet 5.562.3%Opus 5.556.3%Fable 5.1020406080Oct 202669.9%Fable 5.261.2%Gemini 4 Carbon55.0%Gemini 4 Argon62.3%Opus 5.556.3%Fable 5.1

Settings and run setups

HarnessProviders
Benchmark's ownAnthropic
Not statedGoogle

Download image

Results

ModelScoreSource
Claude Fable 5.29 Nov 202669.9%Introducing Claude Fable 5.2 and Claude Mythos 5.2Corroborated The same number is printed in 1 other lab document: 1Run by Proximal · Proximal's agent harness · highest available effort · mean of 5 trials per task, per the system card · 20-hour budget · safeguards on
Claude Opus 5.522 Sep 202662.3%Introducing Claude Fable 5.2 and Claude Mythos 5.2Corroborated The same number is printed in 3 other lab documents: 1 2 3Run by Proximal · Proximal's agent harness · highest available effort · mean of 5 trials per task, per the system card · 20-hour budget · safeguards on · reported later
Claude Sonnet 5.528 Sep 202661.9%Claude Fable 5.2 and Claude Mythos 5.2 System CardChecked A person checked this number against its source on 10 Nov 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.Run by Proximal · Proximal's agent harness · max reasoning effort · mean of 5 trials per task · 20-hour budget · reported later
Gemini 4 Carbon28 Oct 202661.2%Gemini 4 Carbon: our next frontier modelCorroborated The same number is printed in 1 other lab document: 1Run by Proximal · Proximal's public leaderboardSource image
Claude Fable 5.11 Sep 202656.3%Introducing Claude Fable 5.2 and Claude Mythos 5.2Corroborated The same number is printed in 3 other lab documents: 1 2 3Run by Proximal · Proximal's agent harness · highest available effort · mean of 5 trials per task, per the system card · 20-hour budget · safeguards on · reported later
Gemini 4 Argon30 Sep 202655.0%Gemini 4 Argon: our next era of frontier intelligenceCorroborated The same number is printed in 2 other lab documents: 1 2Run by Proximal · Proximal's public leaderboardSource image

All benchmarks