Head to head
Qwen3 14BvsLlama 3.2 3B Instruct
For tasks, about a month.
Verdict · measured basis
Llama 3.2 3B Instruct costs $0.0001 per solved task against
Qwen3 14B at $0.0012 — ▼ 9.1x cheaper.
Over 200 tasks that is $0.21 a month.
On output token price alone Qwen3 14B looks 1.4x cheaper. The ranking reverses once you divide by pass rate — the cheaper tokens do not win the task.
Verified pricing, pass rate and cost per solved task for Qwen3 14B and Llama 3.2 3B Instruct| Figure | Qwen3 14B | Llama 3.2 3B Instruct |
|---|
| Input $/Mtok | $0.12 | $0.05 |
|---|
| Output $/Mtok | $0.24 | $0.33 |
|---|
| Cached input $/Mtok | $0.12 | $0.05 |
|---|
| Pass rate | 78% | 33% |
|---|
| Cost per solved task | $0.0012measured | $0.0001measured |
|---|
| Cost per month | $0.24 | $0.03 |
|---|
Cost per month against tasks per month · drag the marker
Solid measured · dashed modeled · dotted stale. The gap is the verdict.
Solvency Bench (solvency.dev) — first-party measurement · verified 2026-08-26 · prices from each provider's pricing page; measured rows ignore the tier · Methodology · All models at these settings