Google has expanded its artificial intelligence portfolio with the launch of Gemini 4 Neon, a new AI model designed to handle complex workplace assignments, software engineering and scientific research while keeping operating costs under control.
Announced on November 18, Gemini 4 Neon is positioned as a model that combines advanced reasoning capabilities with faster execution and cost efficiency. The launch marks another step in Google’s efforts to strengthen its position in the increasingly competitive AI market, where companies such as OpenAI and Anthropic are racing to develop models capable of performing increasingly sophisticated tasks.
The new model is designed to sit between Google’s flagship Gemini 4 Carbon and its faster, more economical Gemini 4 Hydrogen. While Carbon is intended for demanding, long-running projects and Hydrogen focuses on everyday tasks requiring quick responses, Neon aims to offer a balance between performance and affordability.
Google says Gemini 4 Neon is particularly suited to enterprise AI applications, including analysing large volumes of information, developing software, processing financial documents, building spreadsheets and supporting research activities. Unlike conventional AI assistants that primarily respond to individual prompts, the model is designed to work through multiple stages of a task, use different tools and produce completed outputs.
One of its key features is a context window of one million tokens, allowing it to process substantial amounts of information in a single interaction. This could prove useful for businesses working with lengthy reports, spreadsheets, PDFs, presentations, audio recordings and videos.
For instance, an analyst could use the model to examine several financial documents, identify patterns and prepare a structured report. Similarly, software developers could use it to understand unfamiliar codebases, identify potential errors and suggest solutions without having to break down every step manually.
The ability to work across different formats is another important feature. Businesses often deal with information spread across multiple files and platforms, making it difficult to connect relevant details. Gemini 4 Neon is designed to bring these inputs together and generate usable outputs, rather than simply summarising individual documents.
Google has also highlighted the model’s ability to create and manipulate complex spreadsheets. In one demonstration, the company used multiple AI agents powered by Neon to recreate an image through a spreadsheet containing around 100,000 cells and more than 500,000 formulas. The exercise demonstrated how the model could use spreadsheet functions to reconstruct visual information without embedding the original image or relying on a separate script.
Performance and competition
Google is positioning Gemini 4 Neon as a strong competitor to Anthropic’s Claude Opus 5.5 and OpenAI’s GPT-6.2 Sol. The company has released benchmark comparisons indicating that its latest model performs competitively across reasoning, coding, business automation and scientific problem-solving.
According to Google’s published evaluations, Gemini 4 Neon scored 61 on the Artificial Analysis Intelligence Index, compared with 59 for Claude Opus 5.5 and 57 for GPT-6.2 Sol. It also recorded a score of 70.9% on the Vals Index, marginally ahead of Claude’s 70.1% and OpenAI’s 66.1%.
The model showed particularly strong results in business automation, scoring 56.3% on AutomationBench, compared with 42.5% for Claude Opus 5.5 and 41.3% for GPT-6.2 Sol.
In scientific and mathematical reasoning, Neon achieved 84% on RiemannBench, ahead of the 80% recorded by both competing models. It also scored 71.3% on HLE-Diamond, a benchmark designed to assess advanced reasoning capabilities.
However, the results are not uniformly in Google’s favour. Claude Opus 5.5 performed slightly better on certain coding evaluations, including Terminal-Bench 4.1, while GPT-6.2 Sol recorded a higher score in reading complex professional charts. The comparisons are based on published benchmark evaluations and should be viewed in the context of the testing methods used.
Pricing and availability
Pricing is another area where Google hopes to attract developers and businesses. Gemini 4 Neon is priced at $3 per million input tokens and $15 per million output tokens. Cached input tokens are charged at $0.15 per million, offering a 95% reduction compared with the standard input rate.
The model also offers a one-million-token context window, making it suitable for applications that require processing large amounts of information. Google says Neon delivered comparable or better overall quality than Gemini 4 Carbon in its early evaluations, while operating at three times the speed and 60% of the cost.
Gemini 4 Neon is being rolled out through the Gemini API in Google AI Studio, Gemini for business and Google’s AI-assisted development platform, Antigravity. It is also available to eligible Google AI Pro and Ultra subscribers through the Gemini app.
With this launch, Google is seeking to make advanced AI more practical for everyday business operations. As organisations increasingly look beyond basic chatbots towards AI agents capable of completing multi-step assignments, the ability to deliver reliable results quickly and at a manageable cost is becoming an important competitive advantage.
The launch of Gemini 4 Neon reflects this shift, with the focus moving from simply generating answers to helping users complete actual work.