Strengths
- It scores 61 on the Artificial Analysis index v5.0.1 at the highest reasoning setting (xhigh) and 60 at the default. Gemini 4 Carbon scores 60, Claude Opus 5.5 59 and GPT-6.2 Sol 57.
- From 1 December 2026 it costs a quarter less than Carbon and Argon in the API: $3 / $15 against $4 / $20. A task on the Artificial Analysis index cost $6.44 at xhigh and $3.74 at the default setting.
- It is quicker than Carbon: Artificial Analysis measures about 110 tokens per second (Carbon 55), and it finishes a task on their index in 21.5 minutes on average, against 37.1 for Carbon.
- Google says it leads on knowledge-work tests: Vals Index 70.9% (Opus 5.5 70.1%), AutomationBench 56.3% and HLE-Diamond 71.3%. Google also says it cuts corners and makes up results less than any other Gemini 4 model.
- It takes in about a million tokens of text, images, video, audio and PDFs at once, and in long decode mode, which runs in the background, it can write very long answers. It can operate a computer and check answers against Google Search and Maps.
Weaknesses
- There is no free API tier, and it costs twice as much as Gemini 4 Hydrogen. Until 30 November 2026, Carbon and Argon are cheaper at their introductory $2 / $10.
- At the highest reasoning setting it is wordy and slow to start. It produced 283 million output tokens on the Artificial Analysis test against a 65 million median and sent the first part of its answer after about 57 seconds.
- In Artificial Analysis's independent Terminal-Bench 4.0 run it scores 63.6%, below Carbon (65.2%) and Kimi K3.1 Pro (67.2%). Most of the tests it leads were published by Google itself.
- In the Gemini app you only get it from Google AI Pro at $19.99 a month. The free version and Google AI Plus do not include it.
- Google has not published the full safety report yet. The Neon for cyber version for security work is only for verified experts in the Fairwind Program.
Best for
- demanding analytical work in finance, law and tax (according to Google)
- coding and AI agents that work through long multi-step tasks
- science and maths tasks and work with long documents
- near-Carbon performance in Google AI Pro or at a lower API price
Not ideal for
- high volumes of simple tasks, where Gemini 4 Hydrogen or Gemini 3.5 Flash-Lite is cheaper
- trying out the Gemini API for free
- apps that need an instant answer at high reasoning settings
How to use it in Europe
Since 18 November 2026 you can use it in Europe through the Gemini API and Google AI Studio, but only on the paid tier. In the Gemini app and Google Antigravity it comes with Google AI Pro and Ultra, and companies find it in Gemini Enterprise. On Gemini Enterprise Agent Platform (formerly Vertex AI), Google offers it to verified enterprise customers, including in the eu region, so data can stay in the EU. Under its terms, Google does not train on API data from users in the EEA.
Price compared with similar models
Benchmark results
| Benchmark | Score | Source |
|---|---|---|
| Artificial Analysis Intelligence Index v5.0.1 (xhigh effort) | 61 | artificialanalysis.ai ↗, November 2026 |
| Terminal-Bench 4.0 (Artificial Analysis, xhigh) | 63.6% | artificialanalysis.ai ↗, November 2026 |
| Artificial Analysis output speed (xhigh) | 110 tokens/s | artificialanalysis.ai ↗, November 2026 |
| Vals Index v2.2 | 70.9% (Claude Opus 5.5: 70.1%) | deepmind.google ↗ (vendor-reported), November 2026 |
| Terminal-Bench 4.1, cheating-adjusted | 65.2% (Claude Opus 5.5: 65.8%) | deepmind.google ↗ (vendor-reported), November 2026 |
Benchmarks cover only a narrow slice of ability and vendors often report them at their best settings. Treat them as a rough guide.
More models from this vendor
- Gemini 4 Carbon: Google's most capable model, available since 10 November 2026 in the paid API, for businesses and in Google AI Ultra. It scores 60 on the independent index.
- Gemini 4 Hydrogen: Google's faster model for everyday coding and agents, out since 12 November 2026. It has a free API tier and comes with Google AI Plus and up.
- Gemini 3.5 Flash-Lite: A very fast, low-cost Google model for large volumes of simple tasks, such as pulling data out of documents.
- Gemma 4 31B: Google's largest open model, which you can run on your own computer or server when data must not leave the company.
- Gemma 4 26B A4B: Google's faster open model for your own server. Close to the larger Gemma 4 31B in quality, but quicker to respond.
Similar models from other vendors
- Claude Opus 5.5 (Anthropic): Anthropic's main model for demanding work, coding and long documents, and the one Anthropic recommends starting with.
- DeepSeek V4 Pro (DeepSeek): DeepSeek's largest open-weight model for demanding text tasks at a low price. It does not understand images.
- Muse Wonder (Meta): Meta's strongest model for long-running agents and coding, still released as a preview.
- Mistral Large 4 (Mistral AI): Mistral's largest model, now with open weights. Built for demanding tasks, long documents and agents, with data kept in the EU and a cheaper API than in preview.
- Kimi K3.1 Pro (Moonshot AI): Moonshot AI's newest model, now with open weights, built for agentic coding, office documents and very long outputs.
- Kimi K3 (Moonshot AI): Moonshot AI's open model that topped the open-weight rankings in summer 2026, strong at web browsing and computer use.
Sources
- Google: Introducing Gemini 4 Neon ↗
- Google DeepMind: Gemini 4 Neon model card ↗
- Gemini API: Gemini 4 Neon ↗
- Gemini API pricing ↗
- Gemini API changelog ↗
- Gemini Enterprise Agent Platform: Gemini 4 Neon ↗
- Gemini subscriptions (SK) ↗
- Artificial Analysis: Gemini 4 Neon ↗
- Artificial Analysis: Gemini 4 Neon vs Gemini 4 Carbon ↗
- 9to5Google: Gemini 4 Neon replaces Argon in the Gemini app ↗
- Gemini API Additional Terms of Service ↗