Benchmark
Launches
Mistral AI@mistralai

Introducing Small 3, our most efficient and versatile model yet! Pre-trained and instructed version, Apache 2.0, 24B, 81% MMLU, 150 tok/s. No synthetic data so great base for anything reasoning - happy building! t.co/8WfgzuuwRk

Views
629,784
Likes
3,133
Replies
133
Reposts
450
Quotes
195
Bookmarks
492
Xreply

Introducing Small 3, our most efficient and versatile model yet!

The post announces Small 3 as an efficient and versatile model. It highlights its training format, license, size, benchmark score, speed, lack of synthetic data, and usefulness as a base for reasoning tasks.

Views
629.8K
Likes
3.1K
Replies
133
Reposts
450
Quotes
195
Bookmarks
492
Topics
Model launchOpen-source licensingModel benchmarksDeveloper use casesAI

Analysis

The post announces Small 3 as an efficient and versatile model. It highlights its training format, license, size, benchmark score, speed, lack of synthetic data, and usefulness as a base for reasoning tasks.

Formats
Product LaunchFeature List
Topics
Model launchOpen-source licensingModel benchmarksDeveloper use cases
Categories
TechnologyArtificial intelligenceProduct announcements

Sign in to see the full analysis.

Sign in
Introducing Small 3, our most efficient and versatile model yet! Pre-trained and instructed version, Apache 2.0, 24B, 81% MMLU, 150 tok/s. No synthetic data so great base for anything reasoning - happy building! https://t.co/8WfgzuuwRk
As written
Introducing Small 3, our most efficient and versatile model yet!
From the post
  1. 01
    The Product Hook

    Introduce the new thing and make a clear claim about its strongest qualities.

  2. 02
    The Proof Points

    List the key specifications or credentials in a fast, compact sequence.

  3. 03
    The Use Case

    Explain why the offering is useful and invite people to start using or making things with it.

  4. 04
    The Link

    Provide a direct link for readers who want more information or access.