Test: Gemini 4 Argon is half the price of Claude Opus 5.5 but far less accurate

A head-to-head test of Gemini 4 Argon and Claude Opus 5.5 on three hard tasks found Argon matched Opus on bug hunting at under half the cost, but failed a billing spec test over one bad line and gave wrong answers on every long-context question. Argon costs $2/$10 per million tokens through November 30, rising to $4/$20 on December 1, matching Opus.

• Argon found and fixed all 14 bugs without breaking 93 tests, spending $0.15 versus Opus's $0.34
• On the billing spec, Opus passed 80/80 and 100/100; Argon scored 11/80 and 20/100 due to an extra argument in datetime.combine()
• On long context, Opus named all 33 affected services; Argon was wrong at every document size
• Argon's $2/$10 per million token pricing runs through November 30, then rises to $4/$20