Compare/Llama 3.2 Instruct 3B vs Qwen3.5 2B (Reasoning)

Llama 3.2 Instruct 3BvsQwen3.5 2B (Reasoning)

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
53 tok/s
TTFT
0.35s
Alibaba

Qwen3.5 2B (Reasoning)

Input
$0.02/M
Output
$0.1/M
Speed
357 tok/s
TTFT
0.54s

Winner by Category

Cheaper
Qwen3.5 2B (Reasoning)
Faster (tok/s)
Qwen3.5 2B (Reasoning)
Lower Latency
Llama 3.2 Instruct 3B
Benchmarks (7-5)
Llama 3.2 Instruct 3B

Pricing Comparison

MetricLlama 3.2 Instruct 3BQwen3.5 2B (Reasoning)
Input ($/M tokens)$0.085$0.02
Output ($/M tokens)$0.085$0.1
Cost for 1M input + 100K output tokens:
Llama 3.2 Instruct 3B$0.09
Qwen3.5 2B (Reasoning)$0.03

Speed Comparison

Output Speed (tokens/s) — higher is better
Llama 3.2 Instruct 3B
53 tok/s
Qwen3.5 2B (Reasoning)
357 tok/s
Time to First Token (seconds) — lower is better
Llama 3.2 Instruct 3B
0.35s
Qwen3.5 2B (Reasoning)
0.54s

Benchmark Comparison

Data from Artificial Analysis API — 12 benchmarks

Intelligence Index
9.716.3
Coding Index
3.5
Math Index
3.3
GPQA Diamond
25.5%45.6%
MMLU-Pro
34.7%
LiveCodeBench
8.3%
AIME 2025
3.3%
MATH-500
48.9%
Humanity's Last Exam
5.2%2.1%
SciCode
5.2%2.8%
IFBench
26.2%31.5%
TerminalBench
3.8%
Llama 3.2 Instruct 3B7 wins
5 winsQwen3.5 2B (Reasoning)

Frequently Asked Questions

Which is cheaper, Llama 3.2 Instruct 3B or Qwen3.5 2B (Reasoning)?

Qwen3.5 2B (Reasoning) is cheaper overall. Its blended price (3:1 input/output ratio) is $0.04/M tokens vs $0.09/M for Llama 3.2 Instruct 3B.

Which model performs better on benchmarks?

Llama 3.2 Instruct 3B wins 7 out of 12 benchmarks compared to 5 for Qwen3.5 2B (Reasoning). See the detailed benchmark chart above for per-category results.

Which is faster for real-time applications?

Qwen3.5 2B (Reasoning) generates tokens faster at 357 tok/s vs 53 tok/s. Llama 3.2 Instruct 3B also has lower time-to-first-token (0.35s vs 0.54s).

When should I use Llama 3.2 Instruct 3B vs Qwen3.5 2B (Reasoning)?

Choose based on your priorities: Qwen3.5 2B (Reasoning) for lower cost, Llama 3.2 Instruct 3B for stronger benchmark performance, and Qwen3.5 2B (Reasoning) for faster generation. For latency-sensitive apps, check the TTFT comparison above.