Skip to main content
Back to Explore

Ministral 3 8B

Mistral·Released 2025-12-02
Open SourceMultimodal
Compare

Related Models

Ministral 3 14B2025-12-02Ministral 3 3B2025-12-02Mistral: Ministral 3 14B 25122025-12-02Mistral: Ministral 3 3B 25122025-12-02Mistral: Ministral 3 8B 25122025-12-02

Pricing

Input

$0.15

per 1M tokens

Output

$0.15

per 1M tokens

Blended

$0.15

per 1M tokens

Cheaper than 75% of models. Median price is $0.54/1M tokens.

Cost Calculator

Tokens per day1M
100K100M

Daily

$0.15

Monthly

$4.50

vs. Similar Models

ERNIE 4.5 300B A47BQ:0.0
$0.48+223%
Hermes 4 - Llama-3.1 405B (Reasoning)Q:0.0
$1.50+900%
NVIDIA Nemotron Nano 12B v2 VL (Reasoning)Q:0.0
$0.30+100%
GLM-4.5V (Reasoning)Q:+0.1
$0.90+500%

Performance

83

tokens/sec

Faster than 42% of models

0.47

seconds

Faster than 90% of models

0.47

seconds

Faster than 94% of models

Market Median

94 tok/s

12% slower

Median TTFT

1.11s

58% faster

Throughput/Dollar

552

tok/s per $/1M

Speed Comparison

Grok 4.1 Fast (Non-reasoning)
83 tok/s+0%
Z.ai: GLM 5.1
83 tok/s+0%
GPT-5.5 (high)
83 tok/s+0%

Benchmarks

MMLU-Pro
64.2%
GPQA Diamond
47.1%
HLE
4.3%
LiveCodeBench
30.3%
SciCode
20.8%
TerminalBench Hard
4.5%
MATH-500Not evaluated
AIMENot evaluated
AIME 2025
31.7%
IFBench
29.1%
Long Context Recall
24.0%
Tau2
26.6%
Market AverageTop Score

Open Source

Quick Compare

Similar Models

ERNIE 4.5 300B A47B

Baidu

Q: 9.0$0.48/1M131K ctx
Slower: 71%Pricier: 223%

Solar Pro 2 (Reasoning)

Upstage

Q: 9.0N/A/1M

NVIDIA Nemotron Nano 12B v2 VL (Reasoning)

NVIDIA

Q: 9.0$0.30/1M
Faster: 221%Pricier: 100%

Hermes 4 - Llama-3.1 405B (Reasoning)

Nous Research

Q: 9.0$1.50/1M
Slower: 52%Pricier: 900%

Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)

NVIDIA

Q: 9.1$0.90/1M
Slower: 37%Pricier: 500%

Mistral Large 2 (Nov '24)

Mistral

Q: 9.1$3.00/1M
Slower: 32%Pricier: 1900%
Compare all 7 models