As of: 2026-11-19 22:11 CET. Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/ Canonical page: https://postcutoff.com/m/gemini-4-neon/ # Gemini 4 Neon Google DeepMind, Gemini 4 family. Status: current. Type: reasoning model. Released: 2026-11-18. Training cutoff: not published. ## How to call it | Provider | Model id | Endpoint | Docs | |---|---|---|---| | Gemini API (Google AI Studio, paid tier) | `gemini-4-neon` | https://generativelanguage.googleapis.com/v1beta/models/gemini-4-neon:generateContent | https://ai.google.dev/gemini-api/docs/models | | Gemini app (AI Pro and Ultra, model picker; replaces Argon), Google Antigravity (AI Pro and Ultra), Gemini for business, Android Studio, GitHub Copilot | | | https://gemini.google.com | ## Pricing - Input: $3 per 1M tokens - Output: $15 per 1M tokens - Unit as published: per 1M tokens; cached input $0.15 (95% off) (Google) - Source: https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-neon/ - Last checked: 2026-11-19 ## Specs - Input: text, image, audio, video, pdf - Output: text - Context window: 1,000,000 tokens - Max output: not published - Open weights: no - Licence: proprietary ## What it has not seen - No published training cutoff. Counted from its release (2026-11-18): at least 5 AI events since (1 major, 0 historic), as of 2026-11-19. - Everything since October 2026: https://postcutoff.com/since/2026-10/ ## What stands out - Near-frontier intelligence at mid-tier cost: Artificial Analysis Intelligence Index 61 at xhigh (#3, behind GPT-6.2 Astra and Claude Fable 5.2 at max; Carbon high 60) at $6.44 per index task vs $7.76 for Carbon; ~110 output tokens/s and ~21 min per task vs ~57 tok/s and ~36 min for Carbon. Source: https://artificialanalysis.ai/articles/gemini-4-neon-intelligence-gain - Top agentic-coding score on independent boards: #1 on the Artificial Analysis Coding Agent Index (59 in the Antigravity CLI vs 58 for Claude Opus 5.5 max and Fable 5.2 xhigh) at $6.86 per task vs $12.61 for Opus 5.5; #2 on Arena's Agent Arena Index (1336, 4.2k sessions, $2.82 median cost per task). Source: https://x.com/arena/status/2123169189397709174 ## Notes Sits between Gemini 4 Carbon (hardest long-horizon work) and Hydrogen (fast everyday). Four thinking levels, default high, top xhigh. Vendor numbers: Vals Index 70.9% (#2), Terminal-Bench 4.1 65.2%, FrontierCode 2.0 46.2%, RiemannBench 84.0%, HLE-Diamond 71.3%, SEC-bench Pro 90.5%, 0.0% Gray Swan IPI attack success at highest thinking. Knowledge cutoff not stated in the launch post. Can orchestrate parallel Hydrogen subagents in Antigravity. Fairwind (cyber defenders) access rolling out. Launched 2026-11-18. Model id `gemini-4-neon` per Google's launch post. Model page: https://postcutoff.com/m/gemini-4-neon/