Developers say Gemini 4 Argon hallucinates heavily on real codebases
A roundup of developer complaints, including from Theo, says Argon hallucinates about how systems work and was factually wrong on over 50% of claims in tests, making it unsuitable for coding.
• “Gemini 4 Argon has proven to be outright unusable for real code work in my codebases. It aggressively hallucinates about how systems work.” • “I was genuinely hopeful for Gemini 4 Argon. There are some good parts, but generally speaking, this model should be avoided.” • “This isn’t the usual ‘oh it’s 5% worse in this way, 3% better in others’. In almost every test I ran, Gemini was factually wrong about over 50% of it’s claims. This model is a straight up liability. Keep it away from your code.” • “Just ran Argon on my ‘codebase analysis’ bench. Not looking good. Only model with a >50% false claim rate 🙃” • “It’s still a Gemini model though. Does weird shit regularly and I don’t feel like I can trust it at all.” • “Gemini is last place 100% of the time.” • “Gemini literally got half its score and hallucinated more facts than other models found overall 💀” From Theo https://x.com/theo/status/2110462797440323798?s=61
This post quotes another post: “Google's Philipp Schmid asks for early feedback on Gemini 4 Argon”