Own a mobile shop? Open your free online shop in 5 minutes

Google rolls out Gemini 4 Neon to Pro and Ultra subscribers

2 min read
Google rolls out Gemini 4 Neon to Pro and Ultra subscribers
Gemini 4 Neon brings frontier-level reasoning with faster task completion across coding, knowledge work, and science workflows, now available in the Gemini app for paid subscribers.

What is new

Google DeepMind has introduced Gemini 4 Neon, a new model in the Gemini 4 family that balances depth and efficiency for complex, high-frequency workflows. Neon ranks #3 overall on the Artificial Analysis Intelligence Index at its xhigh thinking level (score of 61), just behind larger, slower models. It also ranks #2 on the Vals Index (70.9%) and #2 on steerability — adapting cleanly to mid-task user corrections — on the Agent Arena Index. The model features a 1 million-token context window and is priced at $3 per million input tokens ($0.15 per million cached, a 95% discount) and $15 per million output tokens.

Frontier performance across enterprise and science

Gemini 4 Neon excels at enterprise work that lives in spreadsheets, PDFs, charts, audio, and video. It can synthesize across all these inputs in a single context window and turn them into structured deliverables. Independent evaluations rank Neon #1 on both the Legal Research Bench and Tax Agent Bench, and #1 on most Surge AI benchmarks across finance, accounting, spreadsheet modeling, and complex constraint following. It scores 73.5% on Chartography for reading dense professional charts and 75.8% on Agents' Last Exam v1.1. On scientific reasoning, Neon leads on Terminal-Bench Science (70.5%), RiemannBench (84.0%), and HLE-Diamond (71.3%).

Production-ready coding and agent workflows

Engineered for long-running agent workflows, Neon sustains performance through complex, tool-heavy tasks end-to-end. It achieves #1 overall on the Artificial Analysis Coding Agent Index (58.6), with 65.2% on Terminal-Bench 4.1 and 46.2% on FrontierCode 2.0 at its default high setting. For production code, Neon handles multi-file debugging and refactoring across repositories with targeted edits. On SWE-sweep, it leads the mini-SWE harness (12.8%) and reaches a state-of-the-art 39.0% in Google Antigravity. It also achieves state-of-the-art on AutoBaxBench hard tier (76%) for secure code generation and 51.26% on WebExploitBench v1.2 for penetration testing at half the cost of competing large models.

Efficiency and cost control

Neon's four thinking levels scale steadily, letting teams dial intelligence and token spend to what each workload needs. At its default high thinking level, Neon pairs a high intelligence score with a 32% lower cost per task than the larger Carbon model. At xhigh thinking, it shows its highest intelligence score while still costing 36%+ less per task than competing models at their highest setting. Neon sits on the Pareto frontier on both the Vals Index and Agent Arena at high thinking level.

Availability

Gemini 4 Neon is rolling out today across multiple surfaces: for developers via the Gemini API paid tier in Google AI Studio and Android Studio; for enterprises via Gemini for business (API and model dropdown); and for consumers — if you're a Google AI Pro or Ultra subscriber, you can select Gemini 4 Neon from the model dropdown in the Gemini app. Google AI Pro and Ultra subscribers can also try Neon in Google Antigravity. For cybersecurity defenders, eligible Fairwind Program members will receive access in the coming days.

Found this useful?

Share it with someone shopping for a phone.

Keep reading

All articles