
Gemini 4 Neon API
gemini-4-neonGemini 4 Neon is Google's newest Gemini 4 model, released 18 Nov 2026 and placed between Carbon and Hydrogen for complex coding, analytical reasoning and production enterprise workflows. Thinking is on by default and bills as output. Call gemini-4-neon at $3 / $15 per 1M tokens, Google's list price; failed requests are not charged.
API Documentation
View complete API reference with all parameters and examples.
Full API Documentation
View complete API reference with streaming, thinking, and more.
Pricing
Billing: Cost = (input_tokens * input_price + output_tokens * output_price) / 1,000,000
Google's List Price
$3 / $15 per 1M tokens, cached input $0.15; one rate across the whole 1M window. No free tier at Google
Gemini 4 Neon Benchmarks
Google: Terminal-Bench 4.1 65.2% (Hydrogen 57%); HLE-Diamond 71.3%; Artificial Analysis Coding Agent Index 58.6, first
Between Carbon and Hydrogen
Google: built for complex coding, analytical reasoning and production enterprise workflows, balancing depth and efficiency
Costs More per Token Here
Carbon and Argon sell 45% below list here and Hydrogen is $0.825 / $4.125; for price-driven calls pick those
1M Context, 65,536-Token Output
1,048,576-token window; text, image, audio, video and PDF input; thinking levels low, medium, high (default), xhigh (native format only)
Thinking On by Default
A one-word reply used 60 thinking tokens in our test; thinking bills as output, so lower the level to spend less
About Gemini 4 Neon
Gemini 4 Neon is a Large Language Model API provided by Google. Gemini 4 Neon is Google's newest Gemini 4 model, released on 18 November 2026. Google describes it as "frontier-level intelligence with fast end-to-end task completion" for software engineering, knowledge work and science, "built for complex coding, analytical reasoning, and production-grade enterprise workflows". In the Gemini 4 line it sits between Carbon, which Google keeps for the hardest long-horizon projects, and Hydrogen, the fast model for everyday execution; Google says Neon "balances depth and efficiency". In the Gemini app it replaces Gemini 4 Argon for Pro and Ultra subscribers. On APIMODELS it costs $3 per 1M input tokens and $15 per 1M output tokens, with cached input at $0.15. That is Google's own list price, with no discount: the reason to call Neon here is one API key and one balance shared with every other model, no Google Cloud project, and access from mainland China. Google lists one rate for the whole 1M-token window, with no long-context surcharge and no scheduled price change. The Gemini 4 Neon benchmarks Google published: 61 on the Artificial Analysis Intelligence Index at xhigh, third overall behind larger, slower models (Hydrogen: 54 at xhigh); 58.6 on the Artificial Analysis Coding Agent Index, first; 70.9% on the Vals Index, second; 1336 on Agent Arena, second; 65.2% on Terminal-Bench 4.1 (Hydrogen: 57%); 70.5% on Terminal-Bench Science; 71.3% on HLE-Diamond (Hydrogen: 58.8%); 84.0% on RiemannBench; 75.8% on Agents' Last Exam v1.1 (Hydrogen: 71.6%); 56.3% on AutomationBench, first (Argon: 51.3%); and 46.2% on FrontierCode 2.0 at the default high level. Google also says Neon has 32% lower cost per task than Carbon at the high level. These are Google's own figures, so run your own prompts before moving a production workload. The price comparison on this site is different from Google's. Carbon and Argon sell here 45% below Google's list ($1.10 / $5.50 through 30 November, $2.20 / $11.00 from 1 December), Hydrogen at $0.825 / $4.125 and Gemini 3.8 Flash at $0.450 / $2.250, while Neon sells at list, so on apimodels.app Neon costs more per token than any of them. Google's 32% figure compares list prices; here, work out cost per task from your own token counts rather than relying on it. Specifications: a 1,048,576-token context window, up to 65,536 output tokens per response, text, image, audio, video and PDF input, and text output. Thinking levels are low, medium, high (the default) and xhigh. On the OpenAI-compatible endpoint you set them with reasoning_effort, where none and minimal run as low and xhigh and max run as high, so xhigh is reachable only in the native Gemini format; there, minimal returns an error. Thinking is on by default: in our pre-launch test a one-word reply used 60 thinking tokens at the default level. Thinking tokens bill at the output rate, so even short answers carry some thinking cost; estimate from total output rather than the visible answer, and lower the thinking level when you want to spend less. As across Gemini 4, Google ignores temperature, top_p and top_k. Cached input bills at $0.15 per 1M, but whether a repeated prefix hits the cache is decided on Google's side, so plan costs as if nothing is cached. We have not yet tested function calling, structured output or search grounding on Neon through APIMODELS. Google offers no free tier for Neon, and the separate "Neon for cyber" variant is limited to Google's Fairwind Program and is not available here. Call it with model gemini-4-neon on the native Gemini endpoint or the OpenAI-compatible /v1/chat/completions, one API key for both. Like any model API it does not succeed on every call, since upstream errors happen; failed requests are not charged. When not to use it: if price per token decides, Gemini 4 Hydrogen ($0.825 / $4.125) or Gemini 3.8 Flash ($0.450 / $2.250) costs far less, and for the hardest long-horizon problems Google points to Gemini 4 Carbon. Through APIMODELS platform, you can access this model via a unified API with transparent pay-as-you-go pricing. Current pricing: Input: $3.00, Output: $15.00 per 1M tokens.
Key Features
- Google's List Price -- $3 / $15 per 1M tokens, cached input $0.15; one rate across the whole 1M window. No free tier at Google
- Gemini 4 Neon Benchmarks -- Google: Terminal-Bench 4.1 65.2% (Hydrogen 57%); HLE-Diamond 71.3%; Artificial Analysis Coding Agent Index 58.6, first
- Between Carbon and Hydrogen -- Google: built for complex coding, analytical reasoning and production enterprise workflows, balancing depth and efficiency
- Costs More per Token Here -- Carbon and Argon sell 45% below list here and Hydrogen is $0.825 / $4.125; for price-driven calls pick those
- 1M Context, 65,536-Token Output -- 1,048,576-token window; text, image, audio, video and PDF input; thinking levels low, medium, high (default), xhigh (native format only)
- Thinking On by Default -- A one-word reply used 60 thinking tokens in our test; thinking bills as output, so lower the level to spend less
Use Cases
Chatbot & Customer Support
Build intelligent conversational systems to automatically answer user queries and improve service efficiency.
Content Generation
Automatically write articles, emails, ad copy, and other text content to boost productivity.
Code Assistant
Assist with code writing, debugging, and code review to accelerate software development.
Data Analysis
Understand and analyze unstructured data, extract key insights, and generate summary reports.
Why APIMODELS
- Unified API -- One API key to access all models, no need to register on multiple platforms
- Transparent Pricing -- Pay-as-you-go with per-call pricing published on every model page, ideal for indie developers and startups
- Instant Access -- Start right after signup; pay by card, PayPal, Alipay or USDT
- Full Documentation -- Detailed API docs with code examples in cURL, Python, and Node.js
Frequently Asked Questions
What is the difference between APIMODELS and the official Google API?
APIMODELS offers the same Gemini 4 Neon model through our aggregation platform. We provide a unified API interface so you do not need separate accounts for each provider - one API key to access all models.
What is Gemini 4 Neon?
Gemini 4 Neon is Google's newest Gemini 4 model, released on 18 November 2026, which Google describes as "frontier-level intelligence with fast end-to-end task completion" for software engineering, knowledge work and science. Google built it for complex coding, analytical reasoning and production-grade enterprise workflows, and places it between Gemini 4 Carbon, kept for the hardest long-horizon projects, and Gemini 4 Hydrogen, the fast model for everyday execution. In the Gemini app it replaces Gemini 4 Argon for Pro and Ultra subscribers. On apimodels.app you call it as gemini-4-neon at $3 / $15 per 1M tokens, the same as Google's list price, with a 1,048,576-token context and up to 65,536 output tokens per response.
Gemini 4 Neon vs Argon, Carbon and Hydrogen: which should I use?
Start with Argon, the comparison most people ask about: in the Gemini app Google has replaced Argon with Neon for Pro and Ultra subscribers, and Google reports 56.3% on AutomationBench for Neon against 51.3% for Argon. Per token, though, Argon is lower here: $1.10 / $5.50 per 1M through 30 November and $2.20 / $11.00 from 1 December, 45% below Google's list, against $3 / $15 for Neon at list. Google places Neon between Carbon, for the hardest long-horizon problems, and Hydrogen ($0.825 / $4.125 here), for fast everyday execution, and reports Neon at 65.2% on Terminal-Bench 4.1 against Hydrogen's 57%. Google also says Neon costs 32% less per task than Carbon at the high level, but that compares list prices; Carbon sells below list here, so measure cost per task on your own prompts. Pick Neon for complex coding agents and analytical or enterprise workflows; pick Hydrogen or Gemini 3.8 Flash ($0.450 / $2.250) when price per token decides, and Carbon for the hardest long-horizon work. All of them share one key and the same endpoints, so switching is the model field.
How much does the Gemini 4 Neon API cost?
$3 per 1M input tokens and $15 per 1M output tokens, with cached input at $0.15 — the same as Google's list price, one rate across the whole 1M window, and no scheduled price change. Thinking tokens bill at the output rate, and at the default high level Neon thinks a lot even on short tasks: in our six auto-graded tasks on 19 November 2026 it used up to 7,397 thinking tokens to write a Chinese paragraph of about 150 characters, and the six tasks cost $0.247 at our prices. With thinking set to low, the same six cost $0.120 and still scored 37/37, so set low (reasoning_effort "low" in the OpenAI format) unless the task is hard. For scale, a request with 10,000 input tokens and 4,000 output tokens (thinking included) costs $0.03 + $0.06 = $0.09. You pay only for the tokens used, failed requests are not charged, and there is no subscription.
What are the Gemini 4 Neon benchmarks?
From Google's launch post: 61 on the Artificial Analysis Intelligence Index at xhigh, third overall behind larger, slower models, and 58.6 on the Artificial Analysis Coding Agent Index, first. Google also reports 65.2% on Terminal-Bench 4.1, 70.5% on Terminal-Bench Science, 71.3% on HLE-Diamond, 84.0% on RiemannBench, 75.8% on Agents' Last Exam v1.1, 56.3% on AutomationBench (first), 70.9% on the Vals Index (second), 1336 on Agent Arena (second) and 46.2% on FrontierCode 2.0 at the default high level. For comparison, Google's figures for Gemini 4 Hydrogen were 57% on Terminal-Bench 4.1 and 58.8% on HLE-Diamond. These are Google's own numbers rather than independent runs, so treat them as a starting point and compare on your own prompts.
Gemini 4 Neon vs Claude Opus 5.5, Claude Fable 5.2 or GPT-6.2 Sol?
Per 1M tokens on apimodels.app, Neon ($3 / $15) sits between Claude Opus 5.5 ($2.40 / $12.00) and Claude Fable 5.2 ($4.80 / $24.00), and costs three times as much as GPT-6.2 Sol ($1.00 / $5.00). All four take about 1M tokens of context, but Neon returns at most 65,536 output tokens per response while the other three go to 128K, which matters for very long generations. Each vendor reports its own benchmarks on its own setup, so those numbers are not a fair head-to-head. Artificial Analysis gives a cross-check (as of 19 November 2026): Neon scores 61 on its Intelligence Index at xhigh, which Google reports as third overall, against 65 for Claude Fable 5.2 at max, and a task on that index cost $6.44 on Neon against $26.54 on Fable 5.2 at list prices. Our head-to-head with Claude Fable 5.2, benchmarks plus our own six-task test, is on the page "Gemini 4 Neon vs Claude Fable 5.2". The reliable test is still your own prompts run through each model here: one key, and switching is the model field.
Is the Gemini 4 Neon API free?
No. Google offers no free tier for Gemini 4 Neon — it is only on the paid tier of the Gemini API — and it bills per token here too. New APIMODELS accounts get $0.10 of trial credit (one new account per IP address every 48 hours), enough for one request with 10,000 input and 4,000 output tokens ($0.09) or a few shorter ones. After that you top up from $10 and pay only for what you use, with no subscription.
How do I call the Gemini 4 Neon API from Python and set the thinking level?
Set the model to gemini-4-neon and authenticate with your apimodels key. With the OpenAI Python SDK: client = OpenAI(base_url='https://api.apimodels.app/v1', api_key='YOUR_KEY'), then client.chat.completions.create(model='gemini-4-neon', reasoning_effort='low', stream=True, messages=[{'role': 'user', 'content': 'Hello'}]). In this OpenAI-compatible format the thinking knob is reasoning_effort: low, medium and high pass straight through, none and minimal are sent as low, and xhigh and max are sent as high. So xhigh is only reachable in the native Gemini format: POST https://api.apimodels.app/v1beta/models/gemini-4-neon:generateContent (:streamGenerateContent for streaming) with a contents array and generationConfig.thinkingConfig.thinkingLevel set to low, medium, high (the default) or xhigh; there, minimal returns an error. Per Google's Gemini 4 guidance, temperature, top_p and top_k are ignored. In our tests low was enough for routine tasks; keep xhigh for hard multi-step problems where high falls short. You do not need a Google Cloud project or billing account, and it works from mainland China.
What are the Gemini 4 Neon specs and limits?
A 1,048,576-token context window and up to 65,536 output tokens per response. Input can be text, image, audio, video or PDF; output is text only, with no image or audio generation. Expect a long wait before the first token: Artificial Analysis measured 40.6 seconds to the first answer token at high and 13.9 seconds at low, and in our tests each short task took 19 to 100 seconds at the default level, so always call it with streaming on (stream: true on /v1/chat/completions, or :streamGenerateContent on the Gemini endpoint) and allow a generous client timeout: a non-streamed request that runs past about 100 seconds is cut off by the connection before the answer arrives; for quick replies, Gemini 4 Hydrogen is Google's fast tier. Cached input bills at $0.15 per 1M, but whether a prefix hits is decided on Google's side, so plan costs as if nothing is cached. We have not yet tested function calling, structured output or search grounding on Neon through APIMODELS, so try them before relying on them in production. The separate "Neon for cyber" variant is limited to Google's Fairwind Program and is not available here. Not every call succeeds, since upstream errors happen, and failed requests are not charged.
What can Gemini 4 Neon do?
Gemini 4 Neon supports: Complex Coding, Enterprise Workflows, 1M Context, Terminal-Bench 4.1 65.2%. See the APIMODELS docs for full parameters and call examples.
Can I use the Gemini 4 Neon API from any country, including mainland China?
Yes. APIMODELS does not restrict access by country: the site and the API answer requests from anywhere, and apimodels.app and api.apimodels.app are reachable directly from mainland China without a proxy. You call Gemini 4 Neon with one APIMODELS key, so you do not need a separate account with the model's developer.
What payment methods are supported? Can I pay with PayPal or crypto?
Yes to both. You can top up with a card (Visa, Mastercard and others, through Stripe), PayPal, Alipay, or USDT on TRON (TRC20) or BNB Smart Chain (BEP20). Card, PayPal and Alipay top-ups are credited immediately; USDT is credited once the transfer confirms on-chain. Auto top-up is optional, and a $50 top-up adds a $1 bonus ($3 on a $100 top-up). The balance is in USD and billed per use, with no subscription.
Related models
Related reading
How to get access, regional availability, and how this model compares with its alternatives.
Comparisons
How this model stacks up against the alternatives on results, price and features.