DeepSeek R1 Distill Llama 70B
deepseek-ai/deepseek-r1-distill-llama-70b
DeepSeek· 131K ctx· mit
open weightsreasoning quant: unknown
spread (output, p90 ÷ p10)
— too few providers
$0.800 – $0.990 · 3 providers · max÷min 1.2×
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 |
|---|---|---|---|---|
Helicone router / aggregator | $0.0300 | $0.130 | — | $0.0391no cache price published |
FastRouter router / aggregator | $0.0300 | $0.140 | — | $0.0400no cache price published |
NanoGPT router / aggregator | $0.700 | $0.700 | $0.350 | $0.477 |
OpenRouter router / aggregator | $0.800 | $0.800 | — | $0.800no cache price published |
Kilo Gateway router / aggregator | $0.800 | $0.800 | — | $0.800no cache price published |
NovitaAI inference host | $0.800 | $0.800 | — | $0.800no cache price published |
Alibaba (China)CN first-party lab | $0.287 | $0.861 | — | $0.339no cache price published |
DigitalOcean inference host | $0.990 | $0.990 | — | $0.990no cache price published |
Data: models.dev (MIT), snapshot 2026-09-20. Prices are vendor-published; verify before purchase. Methodology.