NOUMENOpen app
NOUMEN

celeris · Balanced

Celeris 1

Added Jul 25, 2026

Compare models

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

Loading models…

Use in the app

Celeris 1 is a diffusion language model built for ultra-low-latency classification, extraction, judging, query rewriting, and other short structured responses.

Input
Text
Output
Text
Discovery tier
Balanced

Pricing

USD per 1M tokens
Input
$2.00
Output
$6.00

Reference prices as published by the model provider. API example

Specifications

Context window
8.2K
Max output
8.2K

Side by side

Celeris 1 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…

Celeris 1celerisThis modelPhi 4 MinimicrosoftRemove
Providercelerismicrosoft
Discovery tierBalancedOpen weights
AddedJul 25, 2026Jul 26, 2025
Limits
Context window8.19K128KBest
Max output8.19K16.38KBest
Average output—24 tokens
Pricing, USD per 1M tokens
Input$2.00$0.17Best
Output$6.00$0.68Best
Performance
Request success (24h)—100.0%
Benchmark indexes
Intelligence Index6.36.3
Coding Index14.4Best3.8
Math Index—6.7
Capabilities
Reasoning effort——

Public evaluations

Benchmarks

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

  • Celeris 1
  • Phi 4 Mini

Intelligence Index

  • Celeris 16.3
  • Phi 4 Mini6.3

Celeris 1: Better than 4% of 326 scored models

3.8Catalog models by score55.1

Indexes

  • Coding Index

    Composite of the coding evaluations

    • Celeris 1: 14.4Top
    • Phi 4 Mini: 3.8

  • Math Index

    Composite of the math evaluations

    • Celeris 1: —
    • Phi 4 Mini: 6.7

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Celeris 1: 0.4%
    • Phi 4 Mini: —

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Celeris 1: 0.2%
    • Phi 4 Mini: —

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Celeris 1: 151 Elo
    • Phi 4 Mini: —

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Celeris 1: 301 Elo
    • Phi 4 Mini: —

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Celeris 1: 1.2%
    • Phi 4 Mini: —

  • AA-LCRv1.1

    Reasoning across long documents

    • Celeris 1: 38.0%Top
    • Phi 4 Mini: 15.3%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Celeris 1: 6.8%Top
    • Phi 4 Mini: 4.4%

  • GPQA Diamond

    Graduate-level science questions

    • Celeris 1: 63.1%Top
    • Phi 4 Mini: 33.1%

  • MMLU-Pro

    Multi-subject knowledge and reasoning

    • Celeris 1: 78.0%Top
    • Phi 4 Mini: 46.5%

  • IFBench

    Following precise output instructions

    • Celeris 1: —
    • Phi 4 Mini: 21.1%

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Celeris 1: 0.0%
    • Phi 4 Mini: —

  • SciCode

    Scientific computing in Python

    • Celeris 1: 21.6%
    • Phi 4 Mini: —

  • LiveCodeBench

    Recent competitive programming problems

    • Celeris 1: —
    • Phi 4 Mini: 12.6%

Math

  • AIME 2025

    Competition mathematics, 2025 problems

    • Celeris 1: —
    • Phi 4 Mini: 6.7%

  • AIME

    Competition mathematics

    • Celeris 1: —
    • Phi 4 Mini: 3.0%

  • MATH-500

    Competition math problems

    • Celeris 1: —
    • Phi 4 Mini: 69.6%

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • Celeris 1: —
    • Phi 4 Mini: 0.0%

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Celeris 1: —
    • Phi 4 Mini: 8.2%

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Celeris 1: 38.0%Top
    • Phi 4 Mini: 15.3%

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

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": "celeris-1",
    "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="celeris-1",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)