Note de episod
Google released Gemini 4 Neon on November 18 at $3 per million input tokens and $15 per million output, with a 1 million token context window and four thinking levels. At its default setting it costs 32% less per task than Gemini 4 Carbon and ranks first on the Artificial Analysis Coding Agent Index at 58.6. The same model scored 12.8% on SWE-sweep in a minimal harness and 39.0% inside Google Antigravity. For builders, the lesson is that the harness around a model can matter as much as the model, so test your agent setup before you pay for a bigger one.
Site: ai-nate.com

