Gemini 4 Neon
Gemini 4 Neon is a fast, mid-priced model in the Gemini 4 family from Google, built for coding and agentic work.
Specifications
| Lab | Google DeepMind |
|---|---|
| Context window | 1M tokens |
| Input price | $3.00/1M |
| Output price | $15.00/1M |
| Release | Nov 2026 |
Strengths
- No.1 on the Artificial Analysis Coding Agent Index (58.6); 61 Intelligence Index at xhigh thinking, just behind larger slower models
- 32% lower cost per task than Carbon at default High thinking; four thinking levels from Low to Extra High
- Sustains performance through long tool-heavy agent runs with reliable context compaction and fewer redundant file reads
- Lowest fabrication rate yet in Google Antigravity testing with strong multi-step instruction following
- More secure code than Carbon across AutoBaxBench tiers; state-of-the-art on the hard tier (76%)
Weaknesses
- Flagship-adjacent pricing ($3 in / $15 out per M tokens), roughly 4x Hydrogen for routine work
- Fixed sampling parameters (temperature/top_p/top_k ignored); no explicit context caching or fine-tuning
- New release with limited independent benchmark history beyond launch coverage
Best for
- Repo-level coding agents and long-running development workflows
- Neon-as-planner paired with Hydrogen executors in Antigravity Orchestrator
- Secure-by-default code generation
- Grounded enterprise analysis and STEM reasoning
In Depth: Gemini 4 Neon
Summary
Gemini 4 Neon is an AI model from Google DeepMind.
Released in Nov 2026. It takes audio, file, image, text, and video input and produces text output, with a context window of 1M tokens. Input pricing is $3.00 per 1M tokens and output is $15.00 per 1M tokens on OpenRouter.