NOUMENOpen app
NOUMEN

paperinstruments · Efficient

Ultramarine 1 Flash

Added Oct 15, 2026

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Use in the app

Paper Instruments’ reasoning model for writing, knowledge work, and document tasks. Supports image inputs, tools, structured output, and low, high, or max reasoning effort. Reasoning is always enabled.

  • Reasoning
  • Tools
  • Vision
  • Structured output
  • Parallel tools
Input
Text, Images
Output
Text
Reasoning effort
low · high · max
Discovery tier
Efficient

Pricing

USD per 1M tokens
Input
$0.15
Output
$0.50
Cache read
$0.03

Reference prices as published by the model provider. API example

Specifications

Context window
—
Max output
32.8K

Side by side

Ultramarine 1 Flash compared

Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Ultramarine 1 FlashpaperinstrumentsThis modelMiniMax M3.1 FlashMiniMaxRemove
ProviderpaperinstrumentsMiniMax
Discovery tierEfficientEfficient
AddedOct 15, 2026Oct 19, 2026
Limits
Context window—1.05M
Max output32.77K524.29KBest
Average output—1.4K tokens
Pricing, USD per 1M tokens
Input$0.15$0.15
Output$0.50Best$0.60
Cache read$0.03$0.015Best
Performance
Throughput—266.1 tok/s
Time to first token—3.7 s
Request success (24h)98.2%99.7%Best
Benchmark indexes
Intelligence Index—42.7
Capabilities
ReasoningYesYes
ToolsYesYes
VisionYesYes
Structured outputYesNo
Video inputNoYes
Parallel toolsYesNo
Reasoning effortlow, high, maxlow, medium, high, xhigh, max

Public evaluations

Benchmarks

Independent evaluation results, with how each score ranks among the models in the Noumen catalog.

  • Ultramarine 1 Flash
  • MiniMax M3.1 Flash

Intelligence Index

  • Ultramarine 1 Flash—
  • MiniMax M3.1 Flash42.7
3.8Catalog models by score55.1

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Ultramarine 1 Flash: —
    • MiniMax M3.1 Flash: 72.5%

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Ultramarine 1 Flash: —
    • MiniMax M3.1 Flash: 38.1%

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Ultramarine 1 Flash: —
    • MiniMax M3.1 Flash: 1,473 Elo

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Ultramarine 1 Flash: —
    • MiniMax M3.1 Flash: 1,458 Elo

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Ultramarine 1 Flash: —
    • MiniMax M3.1 Flash: 20.8%

  • AA-LCRv1.1

    Reasoning across long documents

    • Ultramarine 1 Flash: —
    • MiniMax M3.1 Flash: 79.7%

  • MLCR-AA

    Long-context reasoning over medical records

    • Ultramarine 1 Flash: —
    • MiniMax M3.1 Flash: 16.7%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Ultramarine 1 Flash: —
    • MiniMax M3.1 Flash: 48.0%

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Ultramarine 1 Flash: —
    • MiniMax M3.1 Flash: 25.8%

  • SciCode

    Scientific computing in Python

    • Ultramarine 1 Flash: —
    • MiniMax M3.1 Flash: 53.4%

Legacy evaluations

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Ultramarine 1 Flash: —
    • MiniMax M3.1 Flash: 79.7%

Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis

Live routing data

Performance and reliability

Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.

Prompt cache hits
76.6%
About $0.0581 per 1M input tokens with caching
Request success (24h)
98.2%
Reliable

Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.

Hourly request success · last 72 hoursLowest hour 100.0%
  • Healthy (98% or more)
  • Degraded (90–98%)
  • Down (under 90%)
  • No requests

Developers

Use it through the API

Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.

Read the quickstart
curl https://www.noumen.xyz/api/v1/chat/completions \
  -H "Authorization: Bearer $NOUMEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "paperinstruments/ultramarine-1-flash",
    "reasoning_effort": "low",
    "messages": [{ "role": "user", "content": "Explain this stack trace." }]
  }'
import os
from openai import OpenAI

client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
    model="paperinstruments/ultramarine-1-flash",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)