Zidong Taichu 2.0
CASIA · Undisclosed (all-modality, continuous fusion)
The academic all-modality entry: more papers than product
CASIA’s all-modality model, with modality coverage level with WITA-Omni (text/image/video/audio) but productisation at the opposite end of this site: composite 80.1, ranked 24th, no official API, weights under a research licence. Its value is in papers and ablations — public evidence on cross-modal alignment runs deeper than most vendors’.
Last updated Oct 18, 2026Editorial scoring
Bottom line
Eight-dimension breakdown
editorial mapping · 0–100- Zidong Taichu 2.080.1
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 |
|---|---|---|
| 全模态对齐论文 | NeurIPS/AAAI 在投 | Academic channel |
| 权重许可 | 研究许可 | No official API |
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 | 83 · Trained on academic corpora; good on clean receipts | |
| Chart data extraction | 76 · Standard charts | |
| 30-min video QA | 80 · All-modality training; average timing | |
| Mobile GUI control | 66 · Not trained for GUI | |
| Chinese cultural context | 86 · Strong on academic Chinese |
Strengths and limitations
Strengths
- All-modality coverage plus public ablations: the deepest public evidence on cross-modal alignment.
- Chinese context 87: better priors on academic Chinese and policy text than its score peers.
- Research-licensed open weights: methods are reproducible and papers citable.
Limitations
- Research licence: even internal production deployment is excluded — the same wall as LongCat-Next.
- No official API, no SLA: production support is zero.
- Ecosystem 65: tooling targets paper reproduction, not engineering delivery.
Price and real cost
List price
—N/A
No official API; weights under a research licence. 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 Zidong Taichu 2.0
Is Zidong Taichu 2.0 open source?
Licence: Open weights (research licence). Weights are available, but check the exact terms and filing requirements before commercial use.
How much does Zidong Taichu 2.0 cost?
No official API; weights under a research licence. It ranks #25 with a composite of 80.1. Real bills depend on resolution, tiling, caching and volume discounts.
What is Zidong Taichu 2.0 best for?
Cross-modal alignment and all-modality pretraining research; Academic evaluation and method reproduction。
Why does Zidong Taichu 2.0 score 80.1?
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 Visual perception & grounding; its weakest is Ecosystem, cost & availability.
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.