Llama 3.3 70 B Instruct Fp 8 Fast vs Beam 1.0 501 B A23 B: cost comparison
Prices updated 2026-11-07 · computed from live provider data
Short answer: Beam 1.0 501 B A23 B is cheaper on a
3:1 input/output blend — $0.475 per million tokens versus
$0.783, a 39% difference.
Beam 1.0 501 B A23 B offers the larger context window at 1,048,576 tokens, against 24,000 for Llama 3.3 70 B Instruct Fp 8 Fast.
Head to head
| Llama 3.3 70 B Instruct Fp 8 Fast | Beam 1.0 501 B A23 B | |
|---|---|---|
| Input, per 1M tokens | $0.293 | $0.300 |
| Output, per 1M tokens | $2.25 | $1.00 |
| Blended 3:1, per 1M | $0.783 | $0.475 |
| Context window | 24,000 | 1,048,576 |
| Vendor | cloudflare | reflection |
Monthly cost at real volumes
| Workload (input / output) | Llama 3.3 70 B Instruct Fp 8 Fast | Beam 1.0 501 B A23 B |
|---|---|---|
| 1M in / 250K out | $0.856 | $0.550 |
| 10M in / 2.5M out | $8.56 | $5.50 |
| 100M in / 25M out | $85.62 | $55.00 |
Which should you pick?
Pick Beam 1.0 501 B A23 B when you need the longer context window (1,048,576 vs 24,000 tokens). Pick Beam 1.0 501 B A23 B when price per token dominates your workload — high-volume classification, extraction, or batch summarisation.