Qwen 3 14B
qwen/qwen-3-14b
Qwen (Alibaba)· 41K ctx
API onlytool callreasoning quant: unknown
spread (output, p90 ÷ p10)
— too few providers
$0.280 – $0.574 · 2 providers · max÷min 2.0×
2 providers priced · 4 offerings
Cheapest / median / spread use providers that serve the model on their own capacity: first-party lab, hyperscaler, inference host.
Blended $/M assumptions
blended = 90.9% input × [(1 − 70% cache) × input + 70% cache × cache-read] + 9.1% × output + cache writes (30% of uncached input) where a write price is published
A provider that publishes no cache price is blended with cache share 0 (all input at the input price) and labelled “no cache price published” — never penalised against identical in/out prices.
| Provider | Input | Output | Cache rd | Blended |
|---|---|---|---|---|
LLM Gateway router / aggregator | $0.0760 | $0.228 | — | $0.0898no cache price published |
NanoGPT router / aggregator | $0.0800 | $0.240 | $0.0400 | $0.0691 |
Kilo Gateway router / aggregator | $0.120 | $0.240 | — | $0.131no cache price published |
Vercel AI Gateway router / aggregator | $0.120 | $0.240 | — | $0.131no cache price published |
OpenRouter router / aggregator | $0.120 | $0.240 | — | $0.131no cache price published |
SiliconFlow inference host | $0.0700 | $0.280 | — | $0.0891no cache price published |
Alibaba (China)CNfirst-party first-party lab | $0.144 | $0.574 | — | $0.183no cache price published |
Data: models.dev (MIT), snapshot 2026-10-21. Prices are vendor-published; verify before purchase. Methodology.