Head to head
vs QW For light moderate heavy
Verdict · measured basis
Qwen3 VL 32B Instruct costs $0.0001 per solved task against
DeepSeek V4 Flash 0731 at $0.0002 — ▼ 1.8x cheaper .
Over 200 tasks that is $0.02 a month.
On output token price alone DeepSeek V4 Flash 0731 looks 3.5x cheaper. The ranking reverses once you divide by pass rate — the cheaper tokens do not win the task.
Copy link Save scenario Export compare table Planned
Verified pricing, pass rate and cost per solved task for DeepSeek V4 Flash 0731 and Qwen3 VL 32B Instruct Figure DeepSeek V4 Flash 0731 Qwen3 VL 32B Instruct Input $/Mtok $0.06 $0.104 Output $/Mtok $0.12 $0.416 Cached input $/Mtok $0.012 $0.104 Pass rate 89% 67% Cost per solved task $0.0002 measured $0.0001 measured Cost per month $0.04 $0.02
Cost per month against tasks per month · drag the marker
Monthly cost against tasks per month Log-log lines for Qwen3 VL 32B Instruct (measured, $0.0001 per solved task), DeepSeek V4 Flash 0731 (measured, $0.0002 per solved task) from 10 to 100,000 tasks a month. Marker at 200 tasks. 10 100 1k 10k 100k $0.001 $0.01 $0.1 $1 $10 TASKS / MONTH (log) → 200 tasks Qwen3 VL 32B Instruct · MEASURED · $0.0001 per solved task $0.02 Qwen3 VL 32B Instruct MEASURED DeepSeek V4 Flash 0731 · MEASURED · $0.0002 per solved task $0.04 DeepSeek V4 Flash 07… MEASURED Monthly cost against tasks per month Log-log lines for Qwen3 VL 32B Instruct (measured, $0.0001 per solved task), DeepSeek V4 Flash 0731 (measured, $0.0002 per solved task) from 10 to 100,000 tasks a month. Marker at 200 tasks. 10 100 1k 10k 100k $0.001 $0.01 $0.1 $1 $10 TASKS / MONTH (log) → 200 tasks Qwen3 VL 32B Instruct · MEASURED · $0.0001 per solved task $0.02 DeepSeek V4 Flash 0731 · MEASURED · $0.0002 per solved task $0.04 — Qwen3 VL 32B Instruct · MEASURED — DeepSeek V4 Flash 0731 · MEASURED 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