Alibaba · The Agentic Era
Qwen 3
Alibaba's flagship open-weights family. Best non-English capability among open models, strong on coding.
By C.W. Jameson · Published 19 May 2026 · Last reviewed 19 May 2026
Qwen 3 is the open-weights default for any non-English workload. The model ships in sizes from 0.6B to 235B (MoE), each independently competitive with Llama 4 at the same scale. Chinese and Asian-language operators have largely standardised on Qwen.
Field signature
Switches to Chinese mid-output on sufficiently long generations.
Specifications
| Released | 2025 |
|---|---|
| Context window | 128,000 tokens |
| Pricing | Free self-hosted; ~$0.20 - $2 per million tokens via Together / Fireworks |
| Modalities | text · image |
| License | Qwen License (commercial OK with attribution) |
| Era | The Agentic Era |
Strengths
- Multilingual
- Open weights
- Coding
Weaknesses
- English nuance
- Refusal behavior on politically sensitive topics
Authentication markers
The fingerprints by which Qwen 3 can be identified from its output alone.
| Tell | Meaning |
|---|---|
| Occasionally outputs 中文 unprompted. | Qwen training distribution. |
Notable works
- Most-downloaded non-Llama model on Hugging Face
Market position
Free - $2 per million tokens
Partner offer
Partner offerings listed for operator convenience. See disclosure for terms.
View partner →Affiliate link — see disclosure.
Primary sources
- [1] Qwen — Models
From the Almanac shop
The Operator's Compendium
Every agent harness, every routing pattern, every cost trick. 90-page PDF.
$29 — Coming soon