Gemini 4 Argon on QuickSilver Pro
Gemini 4 Argon is Google's new flagship reasoning model, available on QuickSilver Pro the day Google opened the API. It accepts text and images over a 1M-token context. Introductory price: $2.00 input / $10.00 output per million tokens, matching Google's standard endpoints; cached input is $0.10 per million tokens.
At a glance
Reasoning across long documents and images, agent workflows, tools and structured output with 1M context.
Pricing comparison ($/1M tokens)
| Provider | Input | Output | vs QSP |
|---|---|---|---|
| QuickSilver Pro | $2.00 | $10.00 | — |
| Google introductory list price (google/gemini-4-argon) | $2.00 | $10.00 | same |
When to use
Use Gemini 4 Argon for long-context reasoning, image understanding and agent workflows with tools or strict JSON Schema output. It supports tool_choice auto, required and named functions, plus temperature and top_p. Google announced it on September 30, 2026 and opened the API on October 13, 2026.
When to use something else
Reasoning is always on and consumes the total output budget. A verified max_tokens=16 call returned empty content with finish_reason length; max_tokens=64 answered. Allow enough tokens for reasoning and the visible answer. Assistant prefill is unsupported: requests ending with an assistant turn return 400. The pinned Vertex endpoint limits output to 65,536 tokens.
Quickstart (curl)
curl https://api.quicksilverpro.io/v1/chat/completions \
-H "Authorization: Bearer $QSP_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gemini-4-argon",
"messages": [{"role": "user", "content": "Hello!"}]
}'OpenAI-compatible. One-line migration via base_url.
FAQ
Reasoning cannot be disabled: reasoning: {enabled: false} and effort none return 400. Effort low and high work. Reasoning text is returned when present. Reasoning tokens are billed as output and count toward max_tokens, which bounds total output including the visible answer.
The introductory price is $2.00 input / $10.00 output per million tokens, with $0.10 cached input, matching the standard Google endpoints. Google announced $4 / $20 after the introductory window; the current price is not permanent. A verified 46-token prompt plus 41 completion tokens cost $0.000502.
Text and image input, a 1,048,576-token context and up to 65,536 output tokens. tool_choice auto, required and named functions work, as does strict response_format json_schema. temperature and top_p are accepted. Assistant prefill is not supported; end requests with a user or tool turn.
Set base_url=https://api.quicksilverpro.io/v1, use your QSP API key and model="gemini-4-argon" for Chat Completions or Responses. Leave mandatory reasoning enabled. Requests are pinned to standard Google Vertex/global, then Google AI Studio, with no fallback to flex or priority endpoints.