GPT OSS 120 B vs Llama 3.2 11 B Vision Instruct: cost comparison
Prices updated 2026-10-31 · computed from live provider data
Short answer: GPT OSS 120 B is cheaper on a
3:1 input/output blend — $0.093 per million tokens versus
$0.205, a 55% difference.
Both offer the same 128,000-token context window, so cost is the deciding factor.
Head to head
| GPT OSS 120 B | Llama 3.2 11 B Vision Instruct | |
|---|---|---|
| Input, per 1M tokens | $0.047 | $0.049 |
| Output, per 1M tokens | $0.230 | $0.676 |
| Blended 3:1, per 1M | $0.093 | $0.205 |
| Context window | 128,000 | 128,000 |
| Vendor | kyma | cloudflare |
Monthly cost at real volumes
| Workload (input / output) | GPT OSS 120 B | Llama 3.2 11 B Vision Instruct |
|---|---|---|
| 1M in / 250K out | $0.105 | $0.217 |
| 10M in / 2.5M out | $1.05 | $2.17 |
| 100M in / 25M out | $10.49 | $21.75 |
Which should you pick?
Because the context windows are identical, the cheaper option is the better default. Choose GPT OSS 120 B unless you are already standardised on cloudflare tooling and want to avoid a second integration.