Argon uses 62k output tokens per task, more than double GPT-6 Astra

agim𓍯 @aGim_asfOriginally posted 2026.10.15

Open original on X

seeing a lot of gemini subscribers say argon burns through their limits fast the numbers back it up. on artificial analysis's tests, argon used: 62k output tokens per task vs 27k for gpt-6 astra 
110M tokens to run the full test vs 67M for gemini 3.1 pro so the price cut looks great on paper, but argon still costs $1.99 per task now and $3.98 after the intro price ends nov 30 a cheaper token means nothing if the model needs 2x as many cheaper per token ≠ cheaper per task