[KongchangAI]
Product

Darwin ZTC

A decision engine introduced by VIDRAFT that completes judgments within a single forward pass and produces zero output tokens, making it suitable for energy-efficient NPU deployment.

Timeline (last 90 days)

Oct 14

Darwin ZTC scored 72.5 on the SOBACO (Japanese cultural common sense) benchmark, ranking first.

Unverified50%
Oct 14

Among the 27 Japanese benchmarks, Darwin ZTC took 10 first places and reached the top three in 16.

Unverified50%
Oct 14

Darwin ZTC is also first place on the S1MB English track (160 models), with a score of 90.80.

Unverified50%
Oct 14

Darwin ZTC has never received any Japanese-specific training; its Japanese ability comes from cross-lingual transfer.

Unverified50%
Oct 14

In head-to-head matchups on the Japanese track, Darwin ZTC achieved 15 wins, 9 losses, and 3 ties against Perplexity.

Unverified50%
Oct 14

Darwin ZTC scored 88.4 on the JMMLU benchmark, tied for best.

Unverified50%
Oct 14

Darwin ZTC scored 88.0 on the common-sense morality judgment benchmark, ranking first.

Unverified50%
Oct 13

VIDRAFT's Darwin ZTC decision engine completes its judgment in a single forward pass, with zero output tokens.

Unverified50%

All Facts (8)

Source Articles