DeepSeek V3.2 Exp
deepseek-ai/deepseek-v3.2-exp
DeepSeek· 164K ctx· mit
open weights quant: unknown
spread (output, p90 ÷ p10)
— too few providers
$0.410 – $0.431 · 3 providers · max÷min 1.1×
3 providers priced · 3 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 |
|---|---|---|---|---|
ZenMux router / aggregator | $0.220 | $0.330 | — | $0.230no cache price published |
Meganova unclassified | $0.270 | $0.400 | — | $0.282no cache price published |
NovitaAI inference host | $0.270 | $0.410 | — | $0.283no cache price published |
Kilo Gateway router / aggregator | $0.270 | $0.410 | — | $0.283no cache price published |
OpenRouter router / aggregator | $0.270 | $0.410 | — | $0.283no cache price published |
SiliconFlow inference host | $0.270 | $0.410 | — | $0.283no cache price published |
NanoGPT router / aggregator | $0.280 | $0.420 | $0.140 | $0.204 |
Alibaba (China)CN first-party lab | $0.287 | $0.431 | — | $0.300no cache price published |
Data: models.dev (MIT), snapshot 2026-09-20. Prices are vendor-published; verify before purchase. Methodology.