Gemini 4 Argon API: Current Pricing, Access and Carbon Comparison
Gemini 4 Argon has 4 public APIMaster routes from $0.8341424/$4.170712, up to 58.3% off current Google pricing. Compare Carbon, context limits and API setup.
Published · Updated
Gemini 4 Argon API is generally available: Google's release notes mark gemini-4-argon GA on October 13, 2026. This updates the September 30 announcement and Fairwind-first rollout. The identifier is now in Google's official model catalog, with paid Gemini API access and Google AI Ultra access in AI Studio. Argon is unavailable on the Free tier and Google AI Pro plan. Ultra's AI Studio allowance and separately billed API requests are different products.
The Standard API price is $2 input / $10 output per 1M tokens through November 30, 2026, then $4 / $20 from December 1. Implicit cached input costs $0.10, then $0.20 per 1M tokens. The model has a 1,048,576-token shared context; the default synchronous output ceiling is 65,536 tokens, with long decode extending thinking and output through multiple hops within that shared budget.
APIMaster lists 4 public active Argon routes from $0.8341424 / $4.170712 per million input / output tokens: up to ~58.3% off Google's current Standard rate. Carbon is now also generally available, with 4 public active routes from $0.3588624 / $1.794312. Carbon's minimum token quote is about 57.0% lower; cost per completed task still depends on token usage and retries.
| Model ID | Original price: current Google Standard | APIMaster current input / output | Active public routes | Discount | Model card |
|---|---|---|---|---|---|
gemini-4-carbon |
$2 / $10 through 2026-11-30 | From $0.3588624 / $1.794312 | 4 | Up to ~82.1% off | Compare Carbon routes |
gemini-4-argon |
$2 / $10 through 2026-11-30 | From $0.8341424 / $4.170712 | 4 | Up to ~58.3% off | Compare Argon routes |
gemini-3.8-flash |
$0.75 / $3.75 through 2026-12-31 | From $0.30201255 / $1.51006275 | 7 | Up to ~59.7% off | Compare Flash routes |
Checked 2026-11-11 at 10:32 UTC. USD per million input / output tokens. Only public status=1 routes without client-exclusive labels are counted. Carbon ranges from $0.3588624–$2 input / $1.794312–$10 output; Argon from $0.8341424–$2 / $4.170712–$10; 3.8 Flash from $0.30201255–$1.50 / $1.51006275–$7.50. Discounts compare with Google's current Standard promotional rates, not later full prices. The live marketplace card is authoritative.
GPT Trial(GPT-6 Astra, GPT-6 Sol, GPT-5.6, GPT-5.5, gpt-5.6-luna, gpt-5.6-sol, gpt-5.6-terra, gpt-6.1-sol, gpt-6-astra, gpt-6-sol, and gpt-6-luna available)
Sign up and claim a $20 GPT trial
What is Gemini 4 Argon?
Gemini 4 Argon is Google's flagship reasoning model for complex software engineering, mathematics, scientific problem solving, enterprise knowledge work and autonomous agents. It supports text, image, video, audio and PDF inputs and produces text. Code execution, function calling, structured outputs, search grounding, file search and URL context are supported; computer use is a preview capability.
Argon can understand video and help write visualization or simulation code. Its model page explicitly says it does not natively generate images or audio and does not support the Live API. A video-understanding or code-rendering demo should not be read as a native video-generation endpoint. See the Gemini 4 overview for the wider family.
Gemini 4 Argon release date: September announcement, October API launch
The two dates answer different questions:
| Date | Event | What it means for developers |
|---|---|---|
| September 30, 2026 | Initial announcement and limited rollout described in the earlier launch material | Early access and capability information |
| October 13, 2026 | Google Gemini API release notes explicitly mark gemini-4-argon generally available |
A documented stable API model, published prices and token limits |
| November 11, 2026 | Current marketplace check | APIMaster: 4 public active Argon routes, from $0.8341424 / $4.170712; Carbon is also live. |
The earlier article's prediction of an October 8–9 opening window is superseded by the official October 13 GA entry. A Discord tracker entry alone would establish only that a third party detected a name. Here, the release notes and official model page establish the model ID and API launch.
How to access Gemini 4 Argon: paid API and Google AI Ultra
To use Argon through APIMaster, create an APIMaster API key and select gemini-4-argon. This uses APIMaster billing; a Google AI Ultra subscription or a separately billed Google project is not required for this path. The following Google account requirements apply to direct Google API / AI Studio access.
For application development, use a Google project with Cloud Billing enabled, a Gemini API key, a supported region and quota for gemini-4-argon. Google's model page links to Argon in Google AI Studio.
For interactive experimentation in AI Studio, Google's AI plans documentation lists Google AI Ultra access to Argon. Free and Google AI Pro do not include it. Subscription benefits apply to the AI Studio web interface; direct REST or SDK requests use the project's API billing and quotas. An Ultra subscription is not an unlimited API key.
Fairwind remains the cyber-defense program and is expanding to additional partners. It is no longer the only access path described in the current Google documentation. Account quota, region and billing eligibility still determine whether a particular developer can send a request.
Gemini 4 Argon API pricing and introductory end date
Google publishes four consumption modes. The amounts below are USD per 1M tokens; output prices include thinking tokens.
| Mode | Input / output through Nov 30, 2026 | Input / output from Dec 1, 2026 | Implicit cached input: intro / later |
|---|---|---|---|
| Standard | $2 / $10 | $4 / $20 | $0.10 / $0.20 |
| Batch | $1 / $5 | $2 / $10 | $0.05 / $0.10 |
| Flex | $1 / $5 | $2 / $10 | $0.05 / $0.10 |
| Priority | $3.60 / $18 | $7.20 / $36 | $0.18 / $0.36 |
Only implicit caching is supported for Argon; explicit context caching is unsupported. The cached input rate is 95% below uncached input in each listed mode. Tool charges, repeated continuation inputs and thinking tokens can increase the full task cost.
For example, a Standard request billed for 100,000 uncached input tokens and 20,000 output/thinking tokens costs $0.40 during the introductory period, before tools or additional calls. That same token usage costs $0.80 from December 1. Long-decode accounting needs the total usage across all hops, not just the visible final answer.
Gemini 4 Argon is now listed on APIMaster. Checked 2026-11-11: 4 public active routes, starting at $0.8341424 input / $4.170712 output per 1M tokens, up to ~58.3% off Google's current $2 / $10 Standard introductory price (approximately 58.3% savings). Open the gemini-4-argon model card; the live marketplace card is authoritative.
| Public route | Input $/1M | Output $/1M |
|---|---|---|
| Lowest checked | $0.8341424 | $4.170712 |
| Alternative | $0.897156 | $4.48578 |
| Alternative | $0.90078 | $4.5039 |
| Standard-rate route | $2 | $10 |
Snapshot: 2026-11-11; active status=1 routes only. Argon spans $0.8341424–$2 input / $4.170712–$10 output per 1M tokens. The lowest price saves ~58.3% against Google's current $2/$10 Standard price. Marketplace badges compare against the $4/$20 reference that Google schedules for December 1: ~79.1% off that later price, not today's introductory price. Route limits and supported tools vary; Google's native long-decode, caching and background APIs are not automatically guaranteed by a marketplace listing. Check the model card before choosing a route.
Gemini 4 Argon context window and 1M-token output
Google documents 1,048,576 tokens shared across input, thinking and output. This is not a separate 1M input budget plus another 1M output budget.
The model page lists a 65,536-token default synchronous maximum and long decode up to the unified limit. In the Interactions API, output beyond 65,536 requires background execution and a higher max_output_tokens; background execution also requires storage. The server continues in 64K hops, and the application combines each model_output step. The actual available budget is reduced by input and reasoning.
This changes how a coding agent or research workflow should handle results. Poll or reconnect to the same interaction, inspect its terminal status and usage, and allow for several minutes of execution. Raising a synchronous output limit above 65,536 is not equivalent to enabling long decode and can return a 400 error. See Google's long-decode guide for the separate GenerateContent continuation path.
Repeated hop inputs are billed again, with implicit caching discounting repeated tokens. Cancellations and failed interactions still bill work completed before termination. Keep checkpoints, a task budget and cancellation handling for long jobs.
Gemini 4 Argon benchmarks and long-horizon coding
Google presents Argon as a model that carries complex tasks through many steps. Its current DeepMind page includes a large Chromium refactor, Python framework optimization, quantum chemistry optimization and enterprise document workflows. Those are vendor and partner case studies, not reproducible APIMaster tests.
The comparison below preserves figures from the earlier launch screenshot supplied for this article. They are historical launch figures; they should not be confused with a newly run evaluation or a live leaderboard.
| Evaluation | Gemini 4 Argon | GPT-6 Astra | Claude Fable 5.1 | Claude Opus 5.5 |
|---|---|---|---|---|
| Vals Index knowledge work | 68.9% | 63.1% | 65.8% | 67.0% |
| AutomationBench | 51.3% | 41.4% | 31.4% | 42.5% |
| Vals Finance Agent v2 | 65.4% | 53.5% | 58.9% | 58.6% |
| Harvey's Legal Agent Benchmark | 19.6% | 5.4% | 6.7% | 3.8% |
| DeepSWE v1.1 | 77.9% | 74.1% | 67.4% | 74.2% |
| FrontierSWE v2 | 55.0% | 65.5% | 56.3% | 62.3% |
| Vibe Code Bench | 91.9% | 89.6% | 90.3% | 90.3% |
| Terminal-bench 4.0 | 57.4% | 58.2% | 57.9% | 66.4% |
| CWE-bench v1 | 68.0% | 68.0% | 58.0% | 67.0% |
The image is a user-supplied screenshot of a GEEKPark post containing a Google DeepMind announcement graphic. It is an editorial reference, not an independently authenticated original Google post or an APIMaster test.
The table supports a workload-specific comparison. Argon leads DeepSWE and several enterprise rows; Astra leads FrontierSWE, and Opus leads Terminal-bench in this snapshot. Compare actual repository completion, repair turns, tool-call success, source citation quality, latency and total billed cost before selecting a model.
Who announced Argon on X, and what now confirms availability?
The initial announcement discussion involved Google DeepMind, Google, Sundar Pichai and Logan Kilpatrick. Artificial Analysis is an external comparison source.
For current availability, the dated Google API release notes, model specification and pricing page are stronger evidence than an account profile, repost or tracker label. They now independently establish GA status, the exact endpoint, token limits and the introductory pricing end date. This update uses those documents for the access decision and retains the older screenshot only for historical benchmark context.
Gemini 4 Argon on OpenRouter and APIMaster
Gemini 4 Argon is now listed on APIMaster. Checked 2026-11-11: 4 public active routes, starting at $0.8341424 input / $4.170712 output per 1M tokens, up to ~58.3% off Google's current $2 / $10 Standard introductory price (approximately 58.3% savings). Open the gemini-4-argon model card; the live marketplace card is authoritative.
Gemini models and frontier alternatives available on APIMaster
Argon remains an available efficient-model baseline. Carbon is now the latest released Gemini 4 flagship: compare the Carbon vs Argon guide and both live model cards. For routine lower-cost tasks, 3.8 Flash starts at $0.30201255 / $1.51006275 across 7 public active routes, up to about 59.7% off current Google Standard. Prefer 3.8 over the legacy 3.7 label for a new Flash integration; route features vary.
How to use Gemini 4 Argon API, and how to start on APIMaster
For the official Google API, a small text request uses the confirmed model ID:
curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-4-argon:generateContent" \
-H "x-goog-api-key: $GEMINI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"contents": [{"parts": [{"text": "Review this repository migration plan."}]}]
}'
This uses a Google Gemini API key, a paid Google project and Google's billing. It is a basic synchronous example; use the documented background or continuation flow for long decode. It has not been executed against a paid Argon account for this article.
For Gemini 4 Argon through APIMaster, use this basic OpenAI-compatible text request. This example has not been executed as a paid request for this article; advanced Google-native features depend on the selected route.
curl "https://apimaster.ai/v1/chat/completions" \
-H "Authorization: Bearer $APIMASTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gemini-4-argon",
"messages": [{"role": "user", "content": "Review this repository migration plan."}]
}'
Register on APIMaster, create an API key, use the API base, and choose gemini-4-argon from its live model card. Prices start at $0.8341424 / $4.170712 per 1M input / output tokens, with 4 public active routes and up to ~58.3% off Google's current introductory Standard price. Check route restrictions and measure task quality, latency and billed cost.
What the Argon launch changes for developers
Argon remains an available efficient-model baseline. Carbon is now the latest released Gemini 4 flagship: compare the Carbon vs Argon guide and both live model cards. For routine lower-cost tasks, 3.8 Flash starts at $0.30201255 / $1.51006275 across 7 public active routes, up to about 59.7% off current Google Standard. Prefer 3.8 over the legacy 3.7 label for a new Flash integration; route features vary.
FAQ
Is Gemini 4 Argon API released?
Yes. Google's Gemini API release notes mark gemini-4-argon generally available on October 13, 2026. September 30 was the earlier announcement, not the public API GA date.
How do I get access to Gemini 4 Argon?
To use Argon through APIMaster, create an APIMaster API key and select gemini-4-argon. This uses APIMaster billing; a Google AI Ultra subscription or a separately billed Google project is not required for this path. The following Google account requirements apply to direct Google API / AI Studio access.
Is gemini-4-argon an official model ID?
Yes. It is now listed in Google's official model catalog and model specification. A third-party tracker entry alone would not have established that, but the official documentation does.
What is Gemini 4 Argon pricing?
Gemini 4 Argon is now listed on APIMaster. Checked 2026-11-11: 4 public active routes, starting at $0.8341424 input / $4.170712 output per 1M tokens, up to ~58.3% off Google's current $2 / $10 Standard introductory price (approximately 58.3% savings). Open the gemini-4-argon model card; the live marketplace card is authoritative.
Does Gemini 4 Argon have a 1M context window or 1M output?
It has a 1,048,576-token unified budget shared by input, thinking and output. Default synchronous output is limited to 65,536 tokens. Long decode can extend generation within the remaining shared budget; Interactions API long decode needs background execution and a higher max_output_tokens.
Can I use Gemini 4 Argon on OpenRouter or APIMaster?
Gemini 4 Argon is now listed on APIMaster. Checked 2026-11-11: 4 public active routes, starting at $0.8341424 input / $4.170712 output per 1M tokens, up to ~58.3% off Google's current $2 / $10 Standard introductory price (approximately 58.3% savings). Open the gemini-4-argon model card; the live marketplace card is authoritative.
Does Argon beat GPT-6 Astra and Claude Opus 5.5 everywhere?
No. The historical launch table shows different leaders across software engineering, enterprise and terminal tasks. Use the same tools, task fixtures and cost limits for a workload-specific comparison.
Which Gemini model can I use on APIMaster today?
Argon remains an available efficient-model baseline. Carbon is now the latest released Gemini 4 flagship: compare the Carbon vs Argon guide and both live model cards. For routine lower-cost tasks, 3.8 Flash starts at $0.30201255 / $1.51006275 across 7 public active routes, up to about 59.7% off current Google Standard. Prefer 3.8 over the legacy 3.7 label for a new Flash integration; route features vary.
Sources and verification notes
- Google Gemini API release notes — October 13 GA release; October 8 Flash deprecations.
- Gemini 4 Argon model specification — official ID, inputs, capabilities and unified token limits.
- Gemini API pricing — current introductory prices, expiration dates and consumption modes.
- Long decode — background execution, multi-hop output and billing.
- Google AI plans — Ultra access and separate API billing.
- Google DeepMind Gemini — current engineering, science, enterprise and cyber-defense examples.
- OpenRouter public model catalog — Argon listing check on October 14.
- APIMaster Argon marketplace query — public route check on November 11.
Argon specifications were checked against official Google documents at GA; public route prices were refreshed on November 11, 2026. Carbon GA is confirmed by the November 10 changelog. Benchmark figures retained from the launch screenshot are historical attributed data. Pricing, quotas and listings can change; check the linked documents and live marketplace cards before sending traffic.