Qwen 3.6 Plus Preview just landed. It's Alibaba's newest flagship model — and it's available for free on Antbase right now.
This is a proprietary model (no public weights), so you won't find it on most providers. OpenRouter is the only third-party API offering it today, and since Antbase routes through OpenRouter's free tier, you get access at zero cost.
What Makes It Notable
- •1,000,000 token context window — process entire codebases, long documents, or extended conversations without truncation
- •Built-in chain-of-thought reasoning — the model thinks through problems step by step before answering
- •Native function/tool calling — structured tool use out of the box, no prompt engineering required
- •Up to 32K output tokens — longer, more detailed responses when you need them
- •Stronger agentic behavior than Qwen 3.5 — reaches conclusions faster, fewer wasted tokens, more stable output
How It Compares to Qwen 3.5
Qwen 3.5 Plus had a tendency to overthink — long reasoning chains that burned through tokens without adding value. Qwen 3.6 Plus Preview addresses this directly. It's more decisive, reaches conclusions faster, and maintains higher stability across tasks. The hybrid architecture improves efficiency while reducing inference energy consumption.
Early benchmarks suggest it competes with leading state-of-the-art models on reasoning and coding tasks, particularly agentic coding and front-end development.
Where It's Available
Since this is a proprietary model with no downloadable weights, only providers with a direct Alibaba partnership can serve it. Right now that means OpenRouter — and by extension, Antbase.
Provider Status Cost
────────────────────────────────────────────
OpenRouter Available Free (preview)
Antbase (ant:free / ant:auto) Free (via OpenRouter)
Together AI Not available —
DeepInfra Not available —
Fireworks Not available —
SambaNova Not available —
Nebius Not available —Using It on Antbase
You can use Qwen 3.6 Plus Preview in three ways:
1. Direct in the Playground
Open the Playground, select the model from the dropdown or use the button below. No account required for the free tier.
2. Via the API
Point any OpenAI-compatible client at Antbase and specify the model directly:
curl https://api.antbase.ai/v1/chat/completions \
-H "Authorization: Bearer $ANT_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen/qwen3.6-plus-preview:free",
"messages": [{"role": "user", "content": "Explain quantum entanglement in simple terms"}]
}'3. Through ANT Auto Routing
Use ant:auto or ant:free and let the routing engine decide. For reasoning-heavy prompts, Qwen 3.6 Plus Preview will be a strong candidate in the free pool given its 88 quality score and built-in chain-of-thought.
from openai import OpenAI
client = OpenAI(
base_url="https://api.antbase.ai/v1",
api_key="your-ant-api-key",
)
response = client.chat.completions.create(
model="ant:free", # routes to the best free model for your task
messages=[{"role": "user", "content": "Write a recursive Fibonacci in Rust with memoization"}],
)
print(response.choices[0].message.content)Important: It's a Preview
This is a preview release. Alibaba notes that prompts and completions may be used to improve the model. Don't send sensitive data through it. The free tier may also have rate limits depending on OpenRouter availability.
That said — for coding, reasoning, and general-purpose tasks, it's one of the strongest free models available today. A 1M context window with built-in reasoning at zero cost is hard to beat.
Try It Now
Head to the Playground and give it a spin. Drop in a complex coding problem, a long document, or a multi-step reasoning task and see how it handles it.
