Compare/Llama 3.2 Instruct 3B vs Ministral 3 3B

Llama 3.2 Instruct 3BvsMinistral 3 3B

Side-by-side comparison of pricing, 12 benchmarks, and generation speed.

Meta

Llama 3.2 Instruct 3B

Input
$0.085/M
Output
$0.085/M
Speed
50 tok/s
TTFT
0.35s
Mistral

Ministral 3 3B

Input
$0.1/M
Output
$0.1/M
Speed
301 tok/s
TTFT
0.26s

Winner by Category

Cheaper
Llama 3.2 Instruct 3B
Faster (tok/s)
Ministral 3 3B
Lower Latency
Ministral 3 3B
Benchmarks (1-11)
Ministral 3 3B

Pricing Comparison

MetricLlama 3.2 Instruct 3BMinistral 3 3B
Input ($/M tokens)$0.085$0.1
Output ($/M tokens)$0.085$0.1
Cost for 1M input + 100K output tokens:
Llama 3.2 Instruct 3B$0.09
Ministral 3 3B$0.11

Speed Comparison

Output Speed (tokens/s) — higher is better
Llama 3.2 Instruct 3B
50 tok/s
Ministral 3 3B
301 tok/s
Time to First Token (seconds) — lower is better
Llama 3.2 Instruct 3B
0.35s
Ministral 3 3B
0.26s

Benchmark Comparison

Data from Artificial Analysis API — 12 benchmarks

Intelligence Index
9.711.2
Coding Index
4.8
Math Index
3.322.0
GPQA Diamond
25.5%35.8%
MMLU-Pro
34.7%52.4%
LiveCodeBench
8.3%24.7%
AIME 2025
3.3%22.0%
MATH-500
48.9%
Humanity's Last Exam
5.2%5.3%
SciCode
5.2%14.4%
IFBench
26.2%26.8%
TerminalBench
0.0%
Llama 3.2 Instruct 3B1 wins
11 winsMinistral 3 3B

Frequently Asked Questions

Which is cheaper, Llama 3.2 Instruct 3B or Ministral 3 3B?

Llama 3.2 Instruct 3B is cheaper overall. Its blended price (3:1 input/output ratio) is $0.09/M tokens vs $0.10/M for Ministral 3 3B.

Which model performs better on benchmarks?

Ministral 3 3B wins 11 out of 12 benchmarks compared to 1 for Llama 3.2 Instruct 3B. See the detailed benchmark chart above for per-category results.

Which is faster for real-time applications?

Ministral 3 3B generates tokens faster at 301 tok/s vs 50 tok/s. However, Ministral 3 3B has lower time-to-first-token (0.26s vs 0.35s).

When should I use Llama 3.2 Instruct 3B vs Ministral 3 3B?

Choose based on your priorities: Llama 3.2 Instruct 3B for lower cost, Ministral 3 3B for stronger benchmark performance, and Ministral 3 3B for faster generation. For latency-sensitive apps, check the TTFT comparison above.