Llama 3.8 B vs Llama 3.2.3 B Instruct Fast: cost comparison
Prices updated 2026-10-30 · computed from live provider data
Short answer: Llama 3.8 B is cheaper on a
3:1 input/output blend — $0.058 per million tokens versus
$0.113, a 49% difference.
Llama 3.2.3 B Instruct Fast offers the larger context window at 131,072 tokens, against 8,192 for Llama 3.8 B.
Head to head
| Llama 3.8 B | Llama 3.2.3 B Instruct Fast | |
|---|---|---|
| Input, per 1M tokens | $0.050 | $0.050 |
| Output, per 1M tokens | $0.080 | $0.300 |
| Blended 3:1, per 1M | $0.058 | $0.113 |
| Context window | 8,192 | 131,072 |
| Vendor | vercel_ai_gateway | zerogpu |
Monthly cost at real volumes
| Workload (input / output) | Llama 3.8 B | Llama 3.2.3 B Instruct Fast |
|---|---|---|
| 1M in / 250K out | $0.070 | $0.125 |
| 10M in / 2.5M out | $0.700 | $1.25 |
| 100M in / 25M out | $7.00 | $12.50 |
Which should you pick?
Pick Llama 3.2.3 B Instruct Fast when you need the longer context window (131,072 vs 8,192 tokens). Pick Llama 3.8 B when price per token dominates your workload — high-volume classification, extraction, or batch summarisation.