Person

Paddo

An independent developer who conducts production-environment benchmark testing and reproduction evaluations of AI coding models.

Timeline (last 90 days)

Oct 18

Developer Paddo replayed six real changes on a production monorepo, running each model 20 times; Argon passed everything 16 times, versus 13 for Opus.

Unverified50%

Oct 18

In Paddo's test, Argon produced 2.8 times as many output tokens as Opus, with a bill of $58 versus $46.

Unverified50%

All Facts (2)

Unverified

In Paddo's test, Argon produced 2.8 times as many output tokens as Opus, with a bill of $58 versus $46.

50%Unverified

Developer Paddo replayed six real changes on a production monorepo, running each model 20 times; Argon passed everything 16 times, versus 13 for Opus.

50%

Source Articles

Gemini 4 Argon Ties Opus 5.5 — But Costs 50% More Per TaskOct 18