Model
Model explorer

Mistral 7B

OPEN
Mistral AI · Mistral 7B family · released Sep 27, 2023

Debut open 7.3B model; Apache-2.0 weights beat Llama 2 13B on most benchmarks and launched Mistral AI.

ReasoningCodingVisionFunction callingTool useAgentic
172.7
Elo · rank #370
Parameters
7.3B
Active params
7.3B (dense)
Context
8K tokens
Architecture
Dense transformer — 32 layers, grouped-query attention + sliding-window attention
License
Apache 2.0
Languages
—
API price (in/out)
$0.25 / $0.25
Modalities
text
Benchmark results
Bar shows position within the tracked field; marker = field best
ARC-ChallengeReasoning55.5%#106
best: Llama 3.1 405B · 96.9%
ARC-EasyReasoning80.0%#27
best: Phi-3-medium (14B) · 97.7%
BIG-Bench HardReasoning22.0%#137
best: ERNIE 4.5 300B-A47B · 94.3%
GPQA DiamondReasoning5.6%#288
best: GPT-5.6 · 94.6%
GSM8KMath52.2%#130
best: Llama 3.1 405B · 96.8%
HellaSwagReasoning81.3%#70
best: Claude 3 Opus · 95.4%
HumanEvalCoding30.5%#153
best: Claude Opus 4.5 · 99.4%
IFEvalReasoning23.9%#147
best: Gemma 4 26B A4B · 98.5%
MATH-500Math13.1%#188
best: GPT-5 · 99.4%
MBPPCoding47.5%#73
best: Llama-3.3-Nemotron-Super-49B v1 (Reasoning On) · 91.3%
MMLU (EU-21 languages)Knowledge50.6%#9
best: Llama 3.1 70B · 77.1%
MMLU-ProKnowledge22.4%#170
best: Claude Fable 5 · 91.5%
MMLUKnowledge60.1%#205
best: OpenAI o3 · 92.9%
PIQAReasoning83.0%#18
best: GPT-4o mini · 93.1%
TriviaQAKnowledge69.9%#25
best: Sarvam-1 (2B) · 90.6%
WinoGrandeReasoning75.3%#64
best: PaLM 2 · 90.9%
Run it locally
VRAM @ Q4
4.4 GB
VRAM @ FP16
14.6 GB
Fits on (Q4)
RTX 3060 12GBRTX 4070 Ti 16GBRTX 3090 24GBRTX 4090 24GBRTX 5090 32GBM4 Pro 48GBM3 Max 128GBM3 Ultra 512GBA100 80GBH100 80GBH200 141GBB200 192GB
Throughput data unavailable.
Quantizations
GGUF Q4 · AWQ · GPTQ · MLX
Mistral 7B family
Elo progression across releases
Mistral 7BSep 2023172.7
Resources
HFmistralai/Mistral-7B-v0.1GHmistralai/mistral-inferenceDOCSdocs.mistral.ai/models/mistral-7b-0-1
API price $0.25/$0.25 · each benchmark row carries its own source badge (see methodology)