The new GLM 5.3 flash model is performing well on data analysis tasks.
It's finally multimodal, and the performance per $1 is compelling ($0.07/million tokens in, $0.25/million out). Its open-weight, so you can source US-based inference.
Available to try on these demo datasets, live on VerbaGPT Online (app.verbagpt.com):
- Machine Learning (popular datasets)
- MEPS (public healthcare data)
- Weather (ERA5 data slice)
- SEC 13F filings (what money managers are buying/selling, quarterly)