Sci-fi-vy/Mistral-Small-3.2-24B-Instruct-2506

VISIONPricing:Input $0.867 / Output $1.61Concurrent Unit Cost:2Model Size:24BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jun 23, 2025License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Mistral-Small-3.2-24B-Instruct-2506 is a 24 billion parameter instruction-tuned language model developed by Sci-fi-vy, building upon Mistral-Small-3.1. It features a 32768-token context length and is specifically optimized for enhanced instruction following, reduced repetition errors, and a more robust function calling template. This model excels in scenarios requiring precise adherence to instructions and reliable tool use, making it suitable for complex conversational AI and automated task execution.

Loading preview...

Mistral-Small-3.2-24B-Instruct-2506 Overview

Mistral-Small-3.2-24B-Instruct-2506 is an instruction-tuned language model with 24 billion parameters and a 32768-token context length, developed by Sci-fi-vy. It represents an incremental update over its predecessor, Mistral-Small-3.1-24B-Instruct-2503, focusing on key areas of improvement.

Key Capabilities

  • Enhanced Instruction Following: Demonstrates superior ability to follow precise instructions, with an internal accuracy of 84.78% on instruction following tasks, up from 82.75% in the previous version.
  • Reduced Repetition Errors: Significantly decreases instances of infinite generations and repetitive outputs, showing a 2x reduction on challenging prompts (from 2.11% to 1.29%).
  • Robust Function Calling: Features a more reliable and robust function calling template, enabling better integration with external tools and APIs.
  • Multimodal Reasoning: Retains strong vision capabilities, performing well in tasks like ChartQA (87.4%) and DocVQA (94.86%).
  • Improved STEM Performance: Shows slight improvements in STEM benchmarks, including MATH (69.42%), MBPP Plus - Pass@5 (78.33%), and HumanEval Plus - Pass@5 (92.90%).

Good For

  • Complex Instruction-Based Tasks: Ideal for applications requiring strict adherence to detailed user prompts and multi-step instructions.
  • Automated Workflows: Its robust function calling makes it suitable for integrating with external tools and automating tasks.
  • Conversational AI: Benefits from reduced repetition, leading to more natural and coherent dialogue.
  • Multimodal Applications: Can be used for tasks involving both text and image inputs, such as visual question answering and image-based reasoning.

Finetunes
2 models