NOUMENOpen app
NOUMEN

Mistral · Open weights

MS3.2 24B Magnum Diamond

Added Nov 24, 2025Open weights

Compare models

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

Loading models…

Use in the app

rsLoRA finetune of a text-only conversion of Mistral Small 3.2 24B Instruct (2506), built to bring the Magnum mix's Claude-like prose to a smaller footprint. The "Diamond" moniker nods to the extra heat and pressure - pre-tokenization plus custom loss masking - to turn the assistant-tuned base into creative-writing gems, with or without character names or prefill.

Input
Text
Output
Text
Discovery tier
Open weights

Pricing

USD per 1M tokens
Input
$0.49
Output
$0.49

Reference prices as published by the model provider. API example

Specifications

Context window
32.8K
Max output
—
Average output (7 days)
97 tokens
Longer than 9% of 465 models

Side by side

MS3.2 24B Magnum Diamond 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…

MS3.2 24B Magnum DiamondMistralThis modelMistral Code Agent LatestMistralRemove
ProviderMistralMistral
Discovery tierOpen weightsOpen weights
AddedNov 24, 2025Jun 2, 2026
Limits
Context window32.77K262.14KBest
Max output—32.77K
Average output97 tokens316 tokens
Pricing, USD per 1M tokens
Input$0.49$0.40Best
Output$0.49Best$2.00
Performance
Throughput17.3 tok/s—
Time to first token19.8 s—
Request success (24h)59.3%100.0%Best
Benchmark indexes
Intelligence Index—8.6
Coding Index—31.3
Math Index—36.7
Capabilities
ToolsNoYes
Structured outputNoYes
Reasoning effort——

Public evaluations

Benchmarks

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

  • MS3.2 24B Magnum Diamond
  • Mistral Code Agent Latest

Intelligence Index

  • MS3.2 24B Magnum Diamond—
  • Mistral Code Agent Latest8.6
3.8Catalog models by score55.1

Indexes

  • Coding Index

    Composite of the coding evaluations

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 31.3

  • Math Index

    Composite of the math evaluations

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 36.7

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 3.1%

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 0.0%

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 482 Elo

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 544 Elo

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 2.4%

  • AA-LCRv1.1

    Reasoning across long documents

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 32.3%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 3.6%

  • GPQA Diamond

    Graduate-level science questions

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 59.4%

  • MMLU-Pro

    Multi-subject knowledge and reasoning

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 76.2%

  • IFBench

    Following precise output instructions

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 38.1%

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 0.0%

  • SciCode

    Scientific computing in Python

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 32.8%

  • LiveCodeBench

    Recent competitive programming problems

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 44.8%

Math

  • AIME 2025

    Competition mathematics, 2025 problems

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 36.7%

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 18.9%

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 24.9%

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • MS3.2 24B Magnum Diamond: —
    • Mistral Code Agent Latest: 32.3%

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.

Throughput
17.3 tok/s
Faster than 9% of 250 models
Time to first token
19.8 s
Quicker than 2% of 250 models
Request success (24h)
59.3%

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": "Doctor-Shotgun/MS3.2-24B-Magnum-Diamond",
    "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="Doctor-Shotgun/MS3.2-24B-Magnum-Diamond",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

Mistral

Other Mistral models, starting with the same family.