Qwen3.8-27B

Qwen3.8-27B is the compact member of Alibaba’s Qwen3.8 generation — 27 billion parameters, dense, open under Apache 2.0 — and it punches well above its size. Alibaba reports 61.7% on SWE-bench Pro, above its own figure for Claude Opus 4.6, plus 89.2% on GPQA Diamond and 84.3% on the OSWorld computer-use test.

It reads images and video as well as text, holds 262K tokens natively, and thinks by default with adjustable effort. Its design mixes fast linear-attention layers with full attention, which keeps long inputs cheap to process.

For scale: 1,000 requests of about 3,000 tokens in and 1,000 out cost $4.09 — 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.

Qwen3.8-27B 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 Qwen3.8-27B 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. The same model also runs on our own server in Colombo, for work whose data has to stay in the country.

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="qwen3.8-27b",
    messages=[{"role": "user", "content": "Summarise this contract in plain English."}],
)