Ultramarine 1 Flash

TL;DR

Before you orderOffer from Amazon

Get art and craft supplies delivered free with Prime

• Fast, free delivery on millions of items
• Prime Video, Amazon Music and more included
• Member-only deals all year

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

Paper Instruments has introduced Ultramarine 1 Flash, an open model aimed at professional tasks in finance, law, consulting and operations. The company reports a 59.3 score on its PI Knowledge Work benchmark at $0.13 per task, while noting the benchmark is internal and the figures are company-reported.

Paper Instruments has announced Ultramarine 1 Flash, an open model for professional work, and says it scored 59.3 on the company’s internal PI Knowledge Work benchmark at a reported cost of $0.13 per task. The release positions the model for finance, law, consulting and operations, but its performance comparisons come from Paper Instruments’ own evaluation.

Paper Instruments says the model reached a 59.3 benchmark score with maximum reasoning on PI Knowledge Work. The company compares that result with scores and per-task costs for other models and says Ultramarine 1 Flash outscored Claude Fable 5 at maximum reasoning, which it lists at 57.8 and $8.19 per task, and GPT-5.6 Sol at maximum reasoning, listed at 55.1 and $2.09 per task. These are company-reported results, not an independent head-to-head assessment.

The benchmark covers work in finance, law, consulting and operations. According to the description, tasks require finished Excel workbooks, PowerPoint decks and Word documents, as well as work inside business systems such as ERP and MES. Paper Instruments says task-specific rubrics assess substance, accuracy and presentation, and that models run through the same harness.

The company lists Ultramarine API rates of $0.15 per million input tokens, $0.03 for cached input and $0.50 per million output tokens, matching the rates it gives for GLM-5.3 Flash. It also reports that Ultramarine generated 6.5% fewer tokens than GLM-5.3 Flash across six benchmarks, including reasoning tokens. The source says the GLM comparison scores are public, while its token counts come from Paper Instruments’ own runs selected to approximate those scores.

A Lower Reported Cost for Office Tasks

If the reported benchmark results hold up under independent testing, the launch points to a lower-cost option for generating finished business documents and completing structured knowledge-work tasks. Paper Instruments puts Ultramarine at $0.13 per benchmark task, compared with several higher-priced model configurations in its chart. For teams running many such tasks, the claimed cost difference could affect which models are practical to use routinely.

The evidence has limits: the score comes from a benchmark built and described by the company selling the model. Paper Instruments says the benchmark correlates strongly with the Artificial Analysis Intelligence Index, reporting r = 0.81; that is the company’s stated correlation, not a substitute for outside replication or evidence that benchmark performance predicts results in every workplace.

How Paper Instruments Measured Performance

PI Knowledge Work is described as an internal benchmark covering document creation and work in live business systems. Paper Instruments says it uses task-specific rubrics and applies the same harness to each model. The benchmark’s focus is on completed professional deliverables, rather than only short-answer questions.

The company also compares token use across six benchmarks, reporting that Ultramarine 1 Flash generated 6.5% fewer tokens than its base, GLM-5.3 Flash. Its report says the total includes reasoning tokens. The supplied material does not identify all six benchmarks or provide enough detail to independently reproduce the comparisons.

““An open model for professional work.””

— Paper Instruments

Independent Results Are Not Provided

The supplied report does not include an independent replication of the 59.3 score or the cost-per-task comparisons. It also does not specify the announcement date, the full task set, or how scores vary across individual task categories. The benchmark’s reported correlation with another index does not establish that the model will perform equally well across real-world jobs.

Paper Instruments calls the model open, but the source material does not detail the license, the model weights, or any limits on commercial use. Those details matter to organizations evaluating whether they can inspect, adapt or deploy it.

Technical Report and API Details

Paper Instruments directs readers to its technical report for more information about the professional data and evaluations used to train the model. The company also lists an Ultramarine API and a GitHub link. The source material does not give a release schedule or say whether independent benchmark results are planned.

Readers assessing the launch can compare the published methodology and licensing details with their own work tasks, then check whether outside evaluations or practical deployment reports become available.

Key Questions

What is Ultramarine 1 Flash?

Paper Instruments describes it as an open model for professional work in finance, law, consulting and operations.

What score did Paper Instruments report?

The company reports a 59.3 score with maximum reasoning on its internal PI Knowledge Work benchmark.

How much does the model cost?

Paper Instruments lists a benchmark cost of $0.13 per task and API prices of $0.15 per million input tokens, $0.03 per million cached input tokens and $0.50 per million output tokens.

Are the performance comparisons independently verified?

The supplied source contains Paper Instruments’ own benchmark claims and does not provide an independent replication.

Source: hn