Qwen3 30B A3B Instruct 2507
qwen/qwen3-30b-a3b-instruct-2507
spread (output, p90 ÷ p10)
1.0×
$0.300 – $0.500 · 4 providers · max÷min 1.7×
Cheapest / median / spread use providers that serve the model on their own capacity: first-party lab, hyperscaler, inference host.
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 |
|---|---|---|---|---|
Cortecs router / aggregator | $0.0990 | $0.299 | — | $0.117no cache price published |
Ambient inference host | $0.0900 | $0.300 | $0.0900 | $0.109 |
SiliconFlow inference host | $0.0900 | $0.300 | — | $0.109no cache price published |
Kilo Gateway router / aggregator | $0.100 | $0.300 | — | $0.118no cache price published |
Nebius Token Factory inference host | $0.100 | $0.300 | $0.0100 | $0.0711 |
OpenRouter router / aggregator | $0.100 | $0.300 | — | $0.118no cache price published |
NanoGPT router / aggregator | $0.200 | $0.500 | $0.100 | $0.164 |
Clarifai inference host | $0.300 | $0.500 | — | $0.318no cache price published |
Free tiers (rate-limited) · 1
No payment required; rate limits and quotas apply (not tracked yet). Excluded from cheapest, median and spread.
- ModelScope Qwen/Qwen3-30B-A3B-Instruct-2507China-hosted endpoint (modelscope.cn)ModelScope account + access tokenfree tier
Data: models.dev (MIT), snapshot 2026-10-17. Prices are vendor-published; verify before purchase. Methodology.