Grok 4.7 on QuickSilver Pro
Grok 4.7 is xAI's flagship for coding, agentic tasks, and knowledge work, succeeding Grok 4.6. It is built for long-running software engineering and checks its own work before it answers. It accepts text and images across a 500K-token context and supports streaming, tools, and structured output. QuickSilver Pro charges $1.60 input / $4.80 output per million tokens — the published xAI list price, with no QSP markup.
At a glance
Long-running agentic coding and knowledge work with self-verification and a 500K-token context.
Pricing comparison ($/1M tokens)
| Provider | Input | Output | vs QSP |
|---|---|---|---|
| QuickSilver Pro | $1.60 | $4.80 | — |
| xAI list price (Grok API) | $1.60 | $4.80 | same |
When to use
Use Grok 4.7 for long-running software engineering, multi-step agent loops, technical research, and knowledge work where the model should verify its own output before handing it back. It supports both Chat Completions and Responses through the same OpenAI-compatible QSP endpoint, including streaming, function calling, JSON Schema output, and image input.
When to use something else
Reasoning is always on and cannot be disabled — reasoning.enabled=false is rejected; use reasoning_effort to trade depth for latency and cost. For routine chat, extraction, or high-volume automation, a Flash-tier model will usually cost less. For prompts above 200,000 tokens, the whole request bills at $3.20 input and $9.60 output per million tokens, so use retrieval or prompt compaction when the full 500K context is unnecessary.
Quickstart (curl)
curl https://api.quicksilverpro.io/v1/chat/completions \
-H "Authorization: Bearer $QSP_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "grok-4.7",
"messages": [{"role": "user", "content": "Hello!"}]
}'OpenAI-compatible. One-line migration via base_url.
FAQ
For requests up to 200,000 prompt tokens, it costs $1.60 per million input tokens, $4.80 per million output tokens, and $0.40 per million cached-input tokens. Above 200,000 prompt tokens, the whole request bills at $3.20 input, $9.60 output, and $0.80 cached input per million tokens. QuickSilver Pro matches the published xAI rates without a markup.
No. Grok 4.7 always reasons, and a request with reasoning.enabled=false is rejected. Set reasoning_effort to control how deeply it thinks — lower effort is faster and uses fewer output tokens.
Set base_url=https://api.quicksilverpro.io/v1, use your QSP API key, and set model="grok-4.7". Both `/v1/chat/completions` and `/v1/responses` are supported.
Grok 4.7 is xAI's successor to Grok 4.6, tuned for long-running software engineering and for verifying its own work, and it costs 20% less per token: $1.60 / $4.80 versus $2.00 / $6.00. Both keep a 500K-token context with image input, tools, and structured output. Grok 4.6 remains available, so switching is a one-line model change.