tbuy

gemini-4-argon API pricing & access

gemini-4-argon API pricing: $2 per 1M tokens. Gemini 4 Argon is Google’s flagship reasoning model for complex software engineering, math…

gemini-4-argon API access

Gemini 4 Argon is Google’s flagship reasoning model for complex software engineering, mathematical and scientific problem solving, and multi-step agent workflows. It accepts text, images, video, audio and PDF inputs and produces text. Tests with a default API key verified automatic routing, a short stream, a two-turn single-function tool round-trip and test PNG understanding through Google AI Studio (39). The latter three also passed channel-pinned diagnostics through Google Vertex (38); the Vertex results do not verify default automatic routing.

Compare supported API endpoints and capabilities, then access gemini-4-argon through the unified API gateway.

Category: text
Context length: 1048576 tokens
Maximum output: 65536 tokens
Supported APIs: Gemini API, OpenAI Chat Completions API
API endpoints: POST /v1beta/models/gemini-4-argon:generateContent, POST /v1/chat/completions
Input modalities: text, image, video, audio, file
Output modalities: text
Capabilities: reasoning, vision, function calling, tools, caching

gemini-4-argon API pricing

Base input price: $2 per 1M tokens
Base output price: $10 per 1M tokens

Base prices are shown in USD before group-specific adjustments. Open the live pricing page for current access-group prices.

About gemini-4-argon

Gemini 4 Argon

Built for complex coding, mathematical and scientific reasoning, and multi-step agents. Google specifies a unified input/output context of 1,048,576 tokens and a default synchronous output limit of 65,536 tokens. Long decode beyond 64K requires a dedicated execution flow and has not been verified for ordinary platform calls.

The current base price is $2 input and $10 output per million tokens; output billing includes thinking tokens. Google currently lists introductory pricing through November 30, 2026, followed by $4 input and $20 output from December 1. Check the platform catalog and request billing records for actual charges.

Official model details · Official pricing

Use cases and prompting

Access and automatic routing

Model ID: gemini-4-argon. Use an API key with the required model and group permissions and call POST https://modelsok.com/v1/chat/completions. A default key can use the model ID directly with automatic routing; no channel suffix or forced group is required. Access, available channels and charges depend on your account and key configuration and usage records.

Verified channels and scope (October 14, 2026)

Google AI Studio (channel 39): default-key automatic routing. Without pinning a channel or forcing a group, all five requests automatically selected this channel. Short text, a short stream, a two-turn single-function tool round-trip and test PNG understanding passed, with usage and deductions reconciled.
Google Vertex (channel 38): channel-pinned diagnostics. Requests temporarily selected channel 38 without changing the default key configuration or forcing a group. All four requests were served through Vertex; a short stream, a two-turn single-function tool round-trip and test PNG understanding passed, with usage and deductions reconciled. This does not verify that default automatic routing selects Vertex, or the member discount for ordinary Vertex-group requests.

The basic example below still uses default-key automatic routing and includes no channel pin. Check each request’s usage record for its actual channel and charges.

Basic request

curl https://modelsok.com/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gemini-4-argon","messages":[{"role":"user","content":"Hello"}],"max_tokens":1024,"extra_body":{"google":{"thinking_config":{"thinking_level":"low"}}}}'

The example uses the output budget and thinking level from the tests. Thinking and the answer share the output budget; adjust it for your task.

Verified request patterns

Streaming: add "stream": true and "stream_options": {"include_usage": true}. Read incremental content from SSE events. Normal completion, [DONE] and final token usage were verified.
Tool calling: declare functions in tools. The test returned a lookup_inventory call in tool_calls; after the client executed a test function and returned its result with role: "tool" and the matching tool_call_id, the model correctly used the result in its answer. Preserve the full assistant message and its tool-call fields for the next turn, passing them back unchanged through your SDK so additional fields are retained.
Image understanding: include type: "text" and type: "image_url" items in the user message’s content array. The image URL can be data:image/png;base64,.... The model correctly identified text, colors and shapes in the test PNG.

Native Gemini POST /v1beta/models/gemini-4-argon:generateContent is also available. Both the AI Studio automatic-routing tests and the Vertex channel-pinned diagnostics used the OpenAI-compatible endpoint; native Gemini streaming was outside their scope.

Test scope: AI Studio covered one short text response, one short stream, one two-turn single-function round-trip and one test PNG; Vertex covered the latter three. The tool executed only a synthetic inventory lookup, with no real external actions. The image was a 600×300 synthetic PNG used to check text, colors and geometric shapes. Native Gemini streaming, parallel or multiple tools, audio/video, complex documents, very long contexts and dedicated long decode were not verified. These short requests do not establish sustained capacity, an SLA or equivalent verification across other channels.

How to call gemini-4-argon

gemini-4-argon is served through tbuy's unified gateway. Existing OpenAI or Anthropic SDK code keeps working: point base_url at https://modelsok.com and use your tbuy API key.

curl (OpenAI-compatible)

curl https://modelsok.com/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "gemini-4-argon", "messages": [{"role": "user", "content": "Hello"}]}'

Python (openai SDK)

from openai import OpenAI

client = OpenAI(base_url="https://modelsok.com/v1", api_key="YOUR_API_KEY")
response = client.chat.completions.create(
    model="gemini-4-argon",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

Frequently asked questions about gemini-4-argon

What is gemini-4-argon?

How do I call gemini-4-argon?

Create an API key with access to gemini-4-argon, then use the exact model ID and a supported endpoint from the API access section. Request fields depend on the selected endpoint.

How is gemini-4-argon priced?

Pricing depends on the selected provider group and the model billing unit. Refer to the current input, output, request, or media prices in the pricing section.

What is the context window of gemini-4-argon?

The model catalog lists a context window of 1048576 tokens. Check the selected endpoint for request limits.

What is the maximum output of gemini-4-argon?

The model catalog lists a maximum output of 65536 tokens. Your request settings may set a lower limit.

How should I evaluate gemini-4-argon for my project?

Start with the use cases and prompting guidance on this page, then evaluate the model with representative inputs from your project.

Back to AI model API pricing