dots3-note preview
Xiaohongshu · 280B total / 16B active (MoE)
An open model built for image-text notes and speech
An open 280B-A16B model released by Xiaohongshu on 14 August 2026, under Apache-2.0 with 512K context and text, image and speech input — but no video. Agents 93 and documents 90 give a composite of 87.3; generation 75 marks it as an understanding model, not a creator.
Last updated Oct 21, 2026Editorial scoring
Bottom line
Eight-dimension breakdown
editorial mapping · 0–100- dots3-note preview87.3
Weights: perception 20 / documents 18 / charts 14 / video 14 / agents 14 / generation 6 / Chinese 6 / ecosystem 8. Definitions and sources on the methodology page.
Public benchmarks
as reported, not converted| Benchmark / metric | Value | Note |
|---|---|---|
| 参数规模 | 280B-A16B | Official spec |
| 上下文 | 512K | Official spec |
| 许可证 | Apache-2.0 | Commercial use |
Public scores come from different versions and prompts and are not interchangeable with our composite; they are one evidence input.
Scenario mapping
five stress tests, one scale| Scenario | Score | Observation |
|---|---|---|
| Chinese note OCR | 89 · Good on mixed image-text layouts | |
| Chart data extraction | 85 · Fine on standard charts | |
| 30-min video QA | 55 · No video support | |
| Mobile GUI control | 90 · Strong tool calls | |
| Chinese cultural context | 89 · Strong on community context |
Strengths and limitations
Strengths
- 512K context plus agents 93 is a rare combination for long notes and many-image tasks.
- Speech input means image-text-plus-voice scenarios do not need a separate ASR stack.
- Apache-2.0 adds no extra terms for commercial use or fine-tuning.
Limitations
- No video support, which leaves out the core short-video scenario.
- Generation 75: it cannot replace an image generation or editing model.
- Ecosystem 85: public benchmarks and third-party reproduction remain thin.
Price and real cost
List price
¥2/M tokens
≈ ¥2 / 1M output tokens; weights free. For 1,000 images at 1,024px, ~1,100 image tokens each and 300 output tokens, the monthly cost is about ¥1.2 (assuming input at 25% of output price).
Use your volumeAbout dots3-note preview
Is dots3-note preview open source?
Licence: Open weights (Apache-2.0). Weights are available, but check the exact terms and filing requirements before commercial use.
How much does dots3-note preview cost?
≈ ¥2 / 1M output tokens; weights free. It ranks #20 with a composite of 87.3. Real bills depend on resolution, tiling, caching and volume discounts.
What is dots3-note preview best for?
Image-text note understanding and structuring; Speech-plus-image community content analysis; Long-context content moderation。
Why does dots3-note preview score 87.3?
The composite weights eight dimensions: perception 20%, documents 18%, charts 14%, video 14%, agents 14%, generation 6%, Chinese 6%, ecosystem 8%. Its strongest dimension is GUI agents & tool use; its weakest is Generation & editing.
Personal ranking · Not an official score
Every ranking, score and comparison on this page is an independent personal ranking, not an official score, and represents no vendor, institution or regulator. Scores are a subjective mapping of public evidence under our own weights — change the weights and the order changes. Disagree, or spotted a factual error? Contact the site owner: jizhi2024@foxmail.com — we check each report and log corrections in the changelog.