Google's Gemini 4 Neon Rivals Top AI Models At Half The Cost
Google DeepMind is rolling out Gemini 4 Neon, a lower-cost frontier AI model for coding, research, and business tasks, widening access to faster high-end reasoning across Googleβs paid tools. Gemini 4 Neon: Key
Google DeepMind is rolling out Gemini 4 Neon, a lower-cost frontier AI model for coding, research, and business tasks, widening access to faster high-end reasoning across Googleβs paid tools.
Gemini 4 Neon: Key Takeaways
- Gemini 4 Neon starts at $3 per million input tokens and includes a 1 million-token context window
- At its xhigh thinking level, Neon scored 61 on an intelligence index and ranked #3 overall behind larger, slower models
- Neon is rolling out today in the Gemini API, Gemini for business, Google Antigravity, and the Gemini app
- A cyber version will begin reaching eligible Fairwind Program defenders in the coming days
Koray Kavukcuoglu, SVP of Google DeepMind and Googleβs chief AI architect, announced Gemini 4 Neon as a model for complex software engineering, knowledge work, and science workflows.
Pricing starts at $3 per million input tokens, $0.15 per million cached input tokens, and $15 per million output tokens. Neon also has a 1 million-token context window, meaning it can take in very large amounts of material in one session.
Where Neon Fits
Google positions Neon between Gemini 4 Carbon and Gemini 4 Hydrogen. Carbon is for the hardest long-horizon projects, Hydrogen is for fast everyday execution, and Neon is meant to balance depth and efficiency for complex, high-frequency work.
On performance, Google described Neon as a top-tier model across enterprise, analytical, and scientific tasks. It ranked #3 overall with a score of 61 on an intelligence index at its xhigh thinking level, #2 on a benchmark for analytical and enterprise work, and #2 on steerability, which Google defined as adapting cleanly to mid-task user corrections. Google also said Neon recorded its lowest fabrication rate yet in internal Antigravity testing.
That positioning matters because Neon is built to handle the kinds of files office and research work already runs on. Google said it can synthesize spreadsheets, PDFs, charts, audio, and video inside the same 1 million-token window, then turn those inputs into structured deliverables. In Googleβs summary of outside evaluations, Neon hit #1 on legal research and tax work, ranked #1 across most finance, accounting, spreadsheet modeling, and complex constraint-following tests, scored 73.5% on dense chart reading, and reached 75.8% on an agents exam.
Google also tied Neon tightly to coding. The company said it is engineered for long-running agent workflows, cuts down on redundant file reads, shares progress while it works, and handles context compaction reliably in long sessions. Neon ranked #1 overall on a coding agent index with 58.6, scored 65.2% on Terminal-Bench 4.1, and reached 46.2% on FrontierCode 2.0 at its default high setting. On a mini-SWE harness, Google put Neon at 12.8% on its own and 39.0% in Google Antigravity.
Also Read: Googleβs Gemini 4 Argon Matches Rival Costs, Runs Tasks 60% Cheaper
Pricing And Availability
Google is also using Neon to sharpen the pricing ladder inside the Gemini 4 family. Neon has four thinking levels, letting teams trade off intelligence and token spend depending on the workload. At the default high setting, Google said Neon delivers a 32% lower cost per task than Carbon. At high, Google said it also lands on the Pareto frontier on two separate evaluations with roughly 50% to 80% lower cost per task than top competing frontier models. At xhigh, Google said Neon still costs 36%+ less per task than competing models at their highest setting.
Cached input pricing is one of the sharpest cuts. Google lists cached input tokens at $0.15 per million, which it described as a 95% discount. That matters for repeated workflows where the same background material gets reused across many tasks.
| User Group | Gemini 4 Neon Access |
|---|---|
| Developers | Gemini API paid tier in Google AI Studio and Android Studio, plus AI Studio and Google Antigravity for eligible Google AI Pro and Ultra subscribers |
| Enterprises | Gemini for business through the API or model dropdown, and Google Antigravity |
| Consumers | Model dropdown in the Gemini app for Google AI Pro and Ultra subscribers |
| Cybersecurity Defenders | Eligible Fairwind Program members begin receiving access in the coming days |
What To Watch Next
The next test is cybersecurity. Google said Gemini 4 Neon for cyber produces more secure code than Carbon across easy, medium, and hard challenges, and achieved 76% on the hard tier of one security benchmark. It also reached 51.26% on a web exploitation benchmark at half the cost of competing large models, 90.5% on a vulnerability-finding benchmark, and 70.7% on a patching benchmark, which Google said was a 3.4% uplift over Carbon.
Google is also leaning on safety as part of the pitch for long-running agent work. The company said Neon maintains Carbon-level protections against severe misuse and prompt injection while improving alignment and agentic behavior. It matched Carbon with 0.0% cheating on Terminal-Bench 4.0, posted a 0.09% misaligned action rate on Googleβs agentic safety suite, and achieved a 99.2% pass rate on an agent honesty evaluation. With real-time safeguards at its highest thinking level, Google said Neon resisted every attack in one indirect prompt injection benchmark.
The rollout still has limits. Consumer access starts with Google AI Pro and Ultra subscribers, enterprise access runs through Gemini for business, and cyber access begins only with eligible Fairwind Program members. Google did not give a broader date for that cyber rollout.
Read Next: Google Ships Gemini 4 Argon, but Only to Cyber Defenders First
Originally published on Fathom News.
Originally published by Dev.to AI. Aggregated on AIWithGhost for educational purposes β full credit and traffic to the original publisher.