{
  "id": 20698994,
  "title": "Artificial Analysis: Gemini 4 Neon trails only Claude Fable 5.2 and GPT-6.2 Astra on the Intelligence Index and rivals Opus 5.5 on coding at a much lower cost (Artificial Analysis)",
  "url": "https://urgent.news/2026/11/19/artificial-analysis-gemini-4-neon-trails-only-claude-fable-5-2-and",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-19T03:35:01.000Z",
  "source": {
    "name": "Techmeme",
    "slug": "techmeme",
    "url": "https://artificialanalysis.ai/articles/gemini-4-neon-intelligence-gain"
  },
  "original_language": "en",
  "account": null,
  "summary": "Google has announced Gemini 4 Neon, a general-purpose model designed for analytical reasoning, multistep agentic workflows, and complex coding tasks. According to SiliconANGLE, the model is intended to strike a balance between the recently released Gemini 4 Carbon and Gemini 4 Hydrogen. Google stated that Neon completes most tasks roughly three times faster than Carbon at approximately 60% lower costs. Techmeme reports that Neon delivers a small intelligence gain over Carbon at a cost per task that is about 17% lower.\n\nArtificial Analysis research indicates that Neon achieved a score of 61 on the Intelligence Index, ranking third overall behind Claude Fable 5.2 and GPT-6.2 Astra. SiliconANGLE reports that Neon outranked GPT-6.2 Sol and Claude Opus 5.5 on various tests, including science, math, cybersecurity, and multimodal understanding. Regarding coding benchmarks, The New Stack notes that Neon resolved 12.8% of the SWE-sweep benchmark using the mini-SWE harness, though this score increased to 39.0% when run inside Antigravity.\n\nGemini 4 Neon is available via the Gemini app for Ultra and Pro subscribers, Gemini for business, and the Gemini API in Google AI Studio, according to Tech in Asia. Dev.to reports the model uses a transformer architecture with 1.2 trillion parameters and the Neon Engine ASIC to achieve an average latency of 12 ms per 1k tokens.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}