Llama 3.2.3 B Instruct vs Qwen 3 30 B A3 B Fp 8: cost comparison
Prices updated 2026-11-02 · computed from live provider data
Short answer: Qwen 3 30 B A3 B Fp 8 is cheaper on a
3:1 input/output blend — $0.122 per million tokens versus
$0.122, a 0% difference.
Llama 3.2.3 B Instruct offers the larger context window at 131,072 tokens, against 32,768 for Qwen 3 30 B A3 B Fp 8.
Head to head
| Llama 3.2.3 B Instruct | Qwen 3 30 B A3 B Fp 8 | |
|---|---|---|
| Input, per 1M tokens | $0.051 | $0.051 |
| Output, per 1M tokens | $0.335 | $0.335 |
| Blended 3:1, per 1M | $0.122 | $0.122 |
| Context window | 131,072 | 32,768 |
| Vendor | meta-llama | cloudflare |
Monthly cost at real volumes
| Workload (input / output) | Llama 3.2.3 B Instruct | Qwen 3 30 B A3 B Fp 8 |
|---|---|---|
| 1M in / 250K out | $0.135 | $0.135 |
| 10M in / 2.5M out | $1.35 | $1.35 |
| 100M in / 25M out | $13.46 | $13.46 |
Which should you pick?
Pick Llama 3.2.3 B Instruct when you need the longer context window (131,072 vs 32,768 tokens). Pick Qwen 3 30 B A3 B Fp 8 when price per token dominates your workload — high-volume classification, extraction, or batch summarisation.