MiniCPM-o 4.5
ModelBest · 9B dense
A 9B omni-modal edge assistant
A 9B omni-modal model from ModelBest released in May 2026 under Apache-2.0: text, image, video and audio input, ecosystem 92, Chinese 86 and a composite of 76.7. It suits edge speech-plus-vision assistants; documents 78, agents 65 and generation 72 sit only near the usable line.
Last updated Oct 21, 2026Editorial scoring
Bottom line
Eight-dimension breakdown
editorial mapping · 0–100- MiniCPM-o 4.576.7
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 | 9B | Dense model |
| 模态 | 4 | Text/image/video/audio |
| 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 receipt OCR | 78 · Fine on clean receipts | |
| Chart data extraction | 70 · Fine on simple charts | |
| 30-min video QA | 78 · Better on short video and audio | |
| Mobile GUI control | 65 · Not GUI-specific | |
| Chinese cultural context | 86 · Natural everyday context |
Strengths and limitations
Strengths
- Four input modalities at 9B is a rare option for edge speech-plus-vision assistants.
- Ecosystem 92: edge tooling and quantisation scripts are mature.
- Chinese 86 keeps everyday conversation natural.
Limitations
- Documents 78 and charts 70: not enough for complex image-text work.
- Agents 65 and generation 72: neither automation nor creation is a strength.
- Perception 80: fine-grained recognition clearly trails specialists.
Price and real cost
List price
—N/A
Open weights; edge-deployable. 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 MiniCPM-o 4.5
Is MiniCPM-o 4.5 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 MiniCPM-o 4.5 cost?
Open weights; edge-deployable. It ranks #45 with a composite of 76.7. Real bills depend on resolution, tiling, caching and volume discounts.
What is MiniCPM-o 4.5 best for?
Edge speech-plus-vision assistant; Accessibility interaction; Offline multimodal QA。
Why does MiniCPM-o 4.5 score 76.7?
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 Ecosystem, cost & availability; 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.