Mistral AI

Mistral: Mixtral 8x22B Instruct

by Mistral AItext+file->text3 endpoints
Neura Intelligence Index
19.9/ 100
Rank
#290
Confidence
high
Benchmarks
10(26% coverage)

Overview

Mistral's official instruct fine-tuned version of Mixtral 8x22B. It uses 39B active parameters out of 141B, offering unparalleled cost efficiency for its size. Its strengths include: - strong math, coding,...

Capabilities

Text generation
Tool use / Function calling
Structured output
Prompt caching
File input
Top-P sampling
Stop sequences
Deterministic seed

Modalities

Input
TextFile
Output
Text

Technical Specifications

Context Window
65.5K tokens
Max Output
52.4K tokens
Knowledge Cutoff
2024-01-31
Tokenizer
Mistral
Uptime (24h)
10000.0%
Parameters
176B MoE
Blended Cost
$0.9/1M tokens
License
apache-2.0Open Weights
Country
🇫🇷FR
Model Family
Mixtral
Release Date
April 10, 20242y

Supported Parameters

frequency_penaltymax_tokenspresence_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_p

Benchmark Performance

general

BenchmarkScoreSource
Chatbot Arena Elo
1146
aggregated
IFEval
72.4
aggregated
MMLU-Pro
56.8
aggregated

coding

BenchmarkScoreSource
HumanEval
79.3
aggregated

math

BenchmarkScoreSource
GSM8K
90.5
aggregated
MATH
55.2
aggregated

reasoning

BenchmarkScoreSource
ARC-Challenge
91.3
aggregated

language

BenchmarkScoreSource
HellaSwag
88.5
aggregated
WinoGrande
82.1
aggregated

safety

BenchmarkScoreSource
TruthfulQA
55.2
aggregated

Performance Over Time

Pricing

Per 1M tokens
Input$2.00
Output$6.00
Cache read$0.20
Added
April 17, 2024
Last synced 9/18/2026