Emu3.5
BAAI · Undisclosed (native world model)
A world model built on next-state prediction
Trained on 10T+ interleaved vision-language tokens, Emu3.5 approaches Nano Banana on generation/editing and excels at multi-turn consistency. Understanding is not its strength.
Last updated Oct 18, 2026Editorial scoring
Bottom line
Eight-dimension breakdown
editorial mapping · 0–100- Emu3.580.5
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 |
|---|---|---|
| MME 感知 | 781.1 | Low; clearly specialized |
| ImgEdit 多轮 | 4.89 | Above BAGEL 4.45 |
| 训练 token | 10T+ | Interleaved vision-language |
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 receipt OCR | 80 · Not a strength | |
| Chart data extraction | 76 · Weak | |
| 30-min video QA | 75 · Generation-oriented | |
| Mobile GUI control | 66 · Unsupported | |
| Chinese cultural context | 84 · Good visual storytelling context |
Strengths and limitations
Strengths
- Unified discrete tokens make generation and understanding isomorphic.
- Strong multi-turn edit consistency (ImgEdit 4.89).
- Unique long-horizon visual storytelling.
Limitations
- MME perception only 781.1 — weak.
- High inference cost limits generation speed.
- Early productization and APIs.
Price and real cost
List price
—N/A
Open, self-hosted. For 1,000 images at 1,024px, ~1,100 image tokens each and 300 output tokens, the monthly cost is about ¥0.0 (assuming input at 25% of output price).
Use your volumeAbout Emu3.5
Is Emu3.5 open source?
Licence: Open (research license). Weights are available, but check the exact terms and filing requirements before commercial use.
How much does Emu3.5 cost?
Open, self-hosted. It ranks #24 with a composite of 80.5. Real bills depend on resolution, tiling, caching and volume discounts.
What is Emu3.5 best for?
Step-by-step visual guides; Multi-turn image editing; World-model research。
Why does Emu3.5 score 80.5?
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 Generation & editing; its weakest is GUI agents & tool use.
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.