Ultramarine 1 Flash

Paper Instruments' GLM-5.3 Flash post-trained model for finance, law, consulting, and operations deliverables.

Paper Instruments · Enterprise and vertical AI applications (legal, finance, healthcare, sales/marketing, customer support, and similar) · model

Published in Feed

Overview

Paper Instruments announced Ultramarine 1 Flash on October 15, 2026, an open model post-trained from GLM-5.3 Flash to produce finished Excel workbooks, PowerPoint decks, and Word documents and to work inside business systems such as ERP and MES. The company reports first place on JobBench (62.9), FrontierFinance (53.3), and its internal PI Knowledge Work benchmark (59.3 versus 57.8 for Claude Fable 5), and a cost of $0.13 per task against $8.19 for Claude Fable 5 at max effort. These are vendor-reported results. Weights are listed as coming soon on GitHub and Hugging Face, promised within two weeks.

This is a sourced directory entry. Features and capabilities are reported in the linked sources; this entry does not document independent Aymorphic testing.

Key features

  • Post-trained from GLM-5.3 Flash
  • Produces Excel, PowerPoint, and Word deliverables and operates inside ERP and MES systems
  • Vendor-reported: 62.9 JobBench, 53.3 FrontierFinance, 59.3 PI Knowledge Work (internal)
  • Uses 6.5% fewer output tokens than its base model across six benchmarks (vendor-reported)
  • Weights coming soon on GitHub and Hugging Face

Use cases

  • Finance and consulting deliverable drafting (models, decks, memos)
  • Legal and operations document workflows
  • Low-cost agent work inside ERP and manufacturing systems

Pricing

$0.15 per million input tokens, $0.03 per million cached input, $0.50 per million output tokens (same as GLM-5.3 Flash).

Pricing reflects the source record and may change. Confirm current terms with the provider.

Dates and sources

Released
15 Oct 2026
Record updated
18 Oct 2026
Record last checked
18 Oct 2026

Explore the Feed

Other entries filed under enterprise and vertical ai applications (legal, finance, healthcare, sales/marketing, customer support, and similar); inclusion does not imply equivalent features.

Browse the Feed