[Late catch] Ultramarine 1 Flash: a low-cost model for finance, legal and consulting work, weights promised
What is it?
Ultramarine 1 Flash is a model from US startup Paper Instruments, built by applying reinforcement learning to Z.ai's open GLM-5.3 Flash for finance, legal and consulting document work. It trains on tasks that have no automatically checkable answer, graded against per-task rubrics by an LLM judge. It's available now through the company's API, and the weights are promised soon on GitHub and Hugging Face.
Why does it matter?
The company reports 59.3Vendor claim on its own knowledge-work suite at $0.13 per taskVendor claim, far below Claude Fable 5's $8.19Vendor claim. Pricing is $0.15 per million input tokens and $0.50 per million output tokens. Teams that want to run long document work such as diligence memos or litigation chronologies cheaply may want to compare it.
Maturity
PreviewUsable through the API, but the weights aren't out yet. The headline comparison rests on an internal benchmark, so independent measurement is needed.
How to track it
- Weights: check whether the weights and a license actually land on Hugging Face and GitHub.
- Independent scores: look for results on outside evaluations such as artificialanalysis.ai.
- Weak spots: see whether the next version fixes the coding benchmarks where the company's own numbers trail the base model.