MistralJul 19, 2024

Mistral: Mistral Nemo

A 12B parameter model with a 128k token context length built by Mistral in collaboration with NVIDIA. The model is multilingual, supporting English, French, German, Spanish, Italian, Portuguese, Chinese, Japanese, Korean, Arabic, and Hindi. It supports function calling and is released under the Apache 2.0 license.

Context Window
131K
tokens
Max Output
16K
tokens
Released
Jul 19, 2024
Arena Rank

Capabilities

👁Vision
🧠Reasoning
🔧Tool Calling
Prompt Caching
🖥Computer Use
🎨Image Generation

Supported Parameters

Frequency Penalty
Reduce repetition
Max Tokens
Output length limit
min_p
Presence Penalty
Encourage new topics
Repetition Penalty
Penalize repeated tokens
Response Format
JSON mode / structured output
Seed
Deterministic outputs
Stop Sequences
Custom stop tokens
structured_outputs
Temperature
Controls randomness
Tool Choice
Control tool usage
Tool Calling
Function calling support
Top K
Top-K sampling
Top P
Nucleus sampling

Pricing Comparison

RouterInput / 1MOutput / 1MCached Input / 1M
OpenRouter$0.02$0.04
Vercel AI$0.02$0.04
Martian$0.02$0.04

Benchmarks

Open LLM Leaderboard
AverageOpen LLM Leaderboard
24.67/100
IFEvalOpen LLM Leaderboard
63.8/100
BBHOpen LLM Leaderboard
29.68/100
MATH Lvl 5Open LLM Leaderboard
12.69/100
GPQAOpen LLM Leaderboard
5.37/100
MUSROpen LLM Leaderboard
8.48/100
MMLU-PROOpen LLM Leaderboard
27.97/100

Model IDs

OpenRoutermistralai/mistral-nemo

Tags

tool-calling
Compare with another model

Related Models