← Back to all models
M

Mistral Large 3 675B

by Mistral AI
Flagship Free & Open Source 🏆 Ranked #21 of 85
82.1
Overall Score
out of 100
About

Mistral AI's largest open-weight model at 675B parameters, offering frontier performance in an open-source package. Engineered for fast, responsive interactions at scale.

Key Metrics
Context Window
128K
tokens
Avg Response
1200
milliseconds
Input Cost
$2.0
per million tokens
Output Cost
$6.0
per million tokens
Arena ELO
1295
Chatbot Arena rating
MT-Bench
9.1
out of 10
Benchmark Scores
MMLU
88.0%
HumanEval
90.0%
MATH
85.0%
GPQA
60.0%
MT-Bench
91.0/10
Capability Profile
Strengths & Limitations
Strengths
✓ Frontier performance ✓ Open source ✓ Strong reasoning ✓ Multilingual ✓ Large parameter count
Limitations
⚠ Requires massive server infrastructure ⚠ Very large download ⚠ Slow local inference
Ideal Use Cases
Enterprise AI Research Complex reasoning Multilingual apps Fine-tuning base
Model Details
Provider Mistral AI
Released 2025-05-01
Type Free & Open Source
Multimodal No
Tier Flagship
Global rank #21 / 85
Pricing (USD)
Input tokens $2.0/M
Output tokens $6.0/M
Per 1,000 tokens ≈ $0.0020 input / $0.0060 output
All Benchmarks
MMLU 88.0%
HumanEval 90.0%
MATH 85.0%
GPQA 60.0%
MT-Bench 9.1/10
Arena ELO 1295
Compare this model View Rankings

You might also consider