Qwen3.8-Max

Qwen3.8-Max is Alibaba’s top model and the first Max-class Qwen whose weights you can download — a 2.4-trillion-parameter mixture of experts. Alibaba trained it to check its own visual output, and reports 67.7% on SWE-bench Pro, 92.6% on GPQA Diamond and 86.1% on the OSWorld computer-use test.

Through the API it reads images and video and holds a million tokens; the downloadable version is text-only with a shorter context. It thinks before answering, at three effort levels.

For scale: 1,000 requests of about 3,000 tokens in and 1,000 out cost $13.26 — 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.

Qwen3.8-Max is billed in rupees, on the same monthly invoice as every other model your team uses — no foreign card, no separate Alibaba account, and one key for all of it.

Prompts sent to Qwen3.8-Max 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.

Alibaba says the Qwen family covers 201 languages and dialects, without naming Sinhala or Tamil or publishing results for either. Test it on your own messages first.

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="qwen3.8-max",
    messages=[{"role": "user", "content": "Summarise this contract in plain English."}],
)