Google Releases Gemini 4 Argon API With 1M Token Output Limit

- Google’s Gemini 4 Argon is here, spitting out 1M tokens per response. It’s a sparse MoE model, not open source, and it costs $10 just to output that much text. Google claims it leads the Vals Index at 68.90%, beating rivals on long-context tasks like legal and code. But let’s be real: if your AI needs a million words to tell you Bitcoin is going up, it’s useless. The model hallucinates, times out, and trails competitors in actual coding benchmarks. It’s expensive, closed, and bloated. Unless you’re paying for enterprise-grade procrastination, skip the API. Your meat wallet will thank you for keeping those hash manifests clean and your attention span intact.