Grok 4.6

Grok 4.6 is xAI’s flagship for code and general work. Its standout number is 94.9% on GPQA Diamond, a test of graduate-level science questions, which independent rankings put first of hundreds of models. It also made a big jump over Grok 4.5 on agent-style office work, and xAI highlights turning a rough description into a working software prototype.

On an independent test of made-up answers, its rate fell from 54% to 34% — better, but still about one wrong answer in three on that test, so keep a check on anything a customer sees. Its context window is 500K tokens, half that of the million-token models.

For scale: 1,000 requests of about 3,000 tokens in and 1,000 out cost $13.80 — a long prompt and a page of answer each. Every rate is published and stays the same from one request to the next. See every model.

Grok 4.6 is billed in rupees, on the same monthly invoice as every other model your team uses — no foreign card, no separate xAI account, and one key for all of it.

Prompts sent to Grok 4.6 are processed outside Sri Lanka. If a workload has to stay in the country, Qwen3.8-27B (Colombo) runs on our own servers in Colombo.

Use the OpenAI SDK you already have — point it at Roar AI and name the model. Read the docs.

from openai import OpenAI

client = OpenAI(base_url="https://api.roar-ai.com/v1", api_key="roar_live_…")

reply = client.chat.completions.create(
    model="grok-4.6",
    messages=[{"role": "user", "content": "Summarise this contract in plain English."}],
)