Google Launches Gemini 4 Neon
Big Tech & Innovation

Google Launches Gemini 4 Neon

Google’s Gemini 4 Neon brings competitive coding scores and lower task costs than Carbon at standard pricing, intensifying the race for enterprise AI customers.

By Benjamin Harper Edited by Michael Foster 3 mins read
Published: Updated:

Google’s Gemini 4 Neon targets coding and enterprise work as AI providers compete on speed and task costs. Its Mountain View headquarters sign is pictured in 2009. Photo: Håkan Dahlström / Wikipedia 

Key Notes

  • Google launched Gemini 4 Neon with competitive coding and enterprise benchmarks, though its comparison table does not show a clean sweep over every rival.
  • API pricing is $3 per million input tokens and $15 per million output tokens, with a one million-token context window.
  • Neon scores 90.5% on SEC-bench Pro, while the separate cybersecurity version is limited to approved Fairwind defenders.

Google has launched Gemini 4 Neon, adding a model for coding, research and business tasks to its AI lineup. Released November 18, Neon combines strong benchmark results with a focus on completing complex assignments faster and at a lower cost than larger models.

The model competes with Anthropic’s Claude Opus 5.5 and OpenAI’s GPT-6.2 Sol. For Google parent Alphabet (NASDAQ: GOOGL), the commercial opportunity is to turn those capabilities into recurring developer and enterprise spending.

Strong Benchmark Results, With Some Exceptions

Google’s comparison table shows Neon leading 16 of 19 listed results, counting two different SWE-sweep configurations separately. Some rival scores are missing, so that does not establish wins over both competing models on every row.

Claude Opus 5.5 remains ahead on Terminal-Bench 4.1 and FrontierCode 2.0, while GPT-6.2 Sol leads on Chartography. Neon’s results also depend on reasoning settings and the tools used to run the tasks.

Artificial Analysis independently ranks Neon third on its Intelligence Index at its highest reasoning setting, with a score of 61. GPT-6.2 Astra and Claude Fable 5.2 rank higher. In Antigravity CLI, Neon is competitive with the leading Claude coding configurations at a substantially lower measured cost per task.

The Cybersecurity Version Has Restricted Access

On SEC-bench Pro, Neon scores 90.5%, compared with 88.34% for GPT-6.2 Sol. The benchmark tests vulnerability discovery and proof-of-concept generation in software systems; its overall leaderboard aggregates 183 V8 and Firefox instances. The score describes performance on those tests, rather than a universal vulnerability-detection rate.

Google separates those capabilities from general availability. Its Neon for cyber version is due to reach approved defenders through the Fairwind Program over the coming days. The general-purpose model has begun rolling out through AI Studio, the Gemini API and other Google products.

Task Costs Matter More Than Token Prices Alone

Neon costs $3 per million input tokens and $15 per million output tokens, with cached inputs priced at $0.15 per million. Its API context window supports one million tokens.

Artificial Analysis measured $6.44 per Intelligence Index task at the highest reasoning setting, about 17% below Carbon’s $7.76 at standard pricing. Neon also completed the evaluator’s tasks about 40% faster. These are benchmark averages, not fixed prices for customer projects.

Carbon’s temporary discount complicates the comparison. Neon’s token prices are 50% above Carbon’s current introductory rates, which end November 30, and 25% below Carbon’s standard rates. Customers choosing between the two need to compare the full cost of finishing their own work.

Alphabet Has to Convert Model Gains Into Revenue

The release follows Google’s introduction of Carbon and expands the choices available to businesses building agents. On Alphabet’s October earnings call, Sundar Pichai said first-party model APIs were processing nearly 32 billion tokens a minute, up 40% from the previous quarter.

Alphabet’s latest results showed the expense of supporting that demand, with capital spending exceeding quarterly operating cash flow. Neon offers Google another way to sell useful AI work at scale. The financial payoff will come from paid usage and the cost of serving it, rather than benchmark rankings alone.

Disclaimer: MarketSpeaker content is provided for informational and educational purposes only and does not constitute financial or investment advice. Investing involves risk, including the possible loss of principal. Verify information independently and consult a qualified professional before making financial decisions.

Big Tech & Innovation, Business, News, Stocks