Using Gemini? Google's new model will change your work
Google and DeepMind have launched Gemini 4 Neon, a new AI model that outperforms Claude Opus 5.5 and GPT-6.2 Sol in leading benchmarks, scoring 61 on Artificial Analysis and 70.9% on the Vals Index. The model excels in agentic coding, information retrieval, scientific challenges, and cybersecurity, with competitive pricing of $3 per million input tokens.
Google and DeepMind have announced the launch of Gemini 4 Neon, a new artificial intelligence model that tops performance charts against Anthropic's Claude Opus 5.5 and OpenAI's GPT-6.2 Sol. The model leads the Artificial Analysis index with a score of 61, and the Vals Index v2.2 with 70.9%. In agentic coding tests (SWE-sweep) it jumped to 39%, compared to 10.1% for Claude and 8.3% for Sol. In information retrieval tests (GraphWalks) it achieved 95.2% accuracy. The model demonstrated practical capabilities: writing a full 3D skiing simulator in the browser, 3D reconstruction of Alan Turing's office, and converting a single image into a Google Sheets spreadsheet with half a million formulas. In cybersecurity, Neon scored 90.5% on SEC-bench Pro and 51.3% on WebExploitBench, and blocked 100% of prompt injection attacks. Pricing is $3 per million input tokens and $15 per million output tokens, with a context window of one million tokens. The model is available to Google AI Pro and Ultra subscribers, in the Antigravity environment, on the business cloud, and in Android Studio.