Tool 04 · compare
Compare lab: overlay three models
Overlaid radar, per-dimension deltas and a spec table. Selection updates instantly, with no reload and no data upload.
Pick 1–3 models3/3
Overlaid capability radar
Eight dimensions, same 0–100 scale
- Qwen3.8-Max95.4
- Doubao Seed 2.0 Pro95.2
- Qwen3.5-397B-A17B94.6
| Dimension | Qwen3.8-Max | Doubao Seed 2.0 Pro | Qwen3.5-397B-A17B | Spread |
|---|---|---|---|---|
| Visual perception & grounding 20% | 97 | 97 | 96 | 1 |
| Document parsing & Chinese OCR 18% | 97 | 96 | 97 | 1 |
| Chart & document reasoning 14% | 96 | 94 | 95 | 2 |
| Video & long-horizon reasoning 14% | 93 | 96 | 92 | 4 |
| GUI agents & tool use 14% | 95 | 93 | 92 | 3 |
| Generation & editing 6% | 90 | 88 | 90 | 2 |
| Chinese context & culture 6% | 95 | 97 | 94 | 3 |
| Ecosystem, cost & availability 8% | 96 | 98 | 98 | 2 |
| Composite | 95.4 | 95.2 | 94.6 | 0.8 |
| Output price | ¥18 | ¥15 | ¥12 | — |
| Context | 1M (256K native) | 128K | 256K (up to 1M) | — |
Up to three models. Red marks the best value per dimension; a spread of 0 means an exact tie.
About the compare lab
Why only three models?
More than three radar lines become hard to read and the table too wide. Run two comparisons, or use the full heatmap.
Are these scores the same as the dossiers?
Yes — same dataset. The tool only visualises; it never recalculates scores.
Can I share a comparison?
This version does not create share links, to avoid encoding selections in URLs. Screenshot it, or cite the model and comparison pages.
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.