Biggest savings on the maker's own price.
Prices checked
Deepseek R1 Distill Qwen 32B at Nscale costs 91% less than Cloudflare Workers AI, the widest of 209 gaps with a reason behind them, checked .
Top 10
all 392 models- 1
Deepseek R1 Distill Qwen 32B
Runs it themselves
Nscale · 91% less than Cloudflare Workers AI · 80k context$0.15/M tokens - 2
Qwen-MT Plus
Resells
Jiekou · 90% less than Alibaba Cloud · 16k context$0.38/M tokens - 3
Qwen3 32B
Resells
OpenRouter · 89% less than Alibaba Cloud · 131k context$0.13/M tokens - 4
Nemotron 3 Super
Resells
NanoGPT · 88% less than OpenRouter · 1M context$0.10/M tokens - 5
Qwen3.8 27B
ResellsNanoGPT · 87% less than Alibaba Cloud · 1M context$0.12/M tokens - 6
Qwen3 Coder 30B A3B Instruct
Resells
Hugging Face Inference Providers · 87% less than Alibaba Cloud · 262k context$0.12/M tokens - 7
Qwen2.5 Coder 32B Instruct
Runs it themselvesNscale · 87% less than OpenRouter · 33k context$0.10/M tokens - 8
Llama 3.2 1B Instruct
Runs it themselves
Inference · 86% less than OpenRouter · 60k context$0.01/M tokens - 9
Qwen3 Coder 480B A35B
Runs it themselves
SiliconFlow · 85% less than Alibaba Cloud · 262k context$0.44/M tokens - 10NanoGPT · 85% less than Alibaba Cloud · 131k context$0.28/M tokens
How we count
Every row here is below the maker's own price with a published reason behind it: the weights are open and somebody else hosts them, a regional price list applies, or the provider says what funds the discount. A price below the maker's that nothing accounts for is listed on the model's page and not ranked, here or elsewhere. 209 models have a reason we can point at, the widest gap being 91%. Prices are per million tokens, blended 3 in : 1 out, at each provider's standard rate.
A price below the maker's is listed only when the provider says why.
- Runs it themselves
- The provider runs the open weights on hardware of its own, so the price is its own rather than a resale.
- Resells
- The provider sells another company's endpoint, at a margin of its own or at the maker's price.
Frequently asked
- What is the cheapest of the prices below the maker's own right now?
- Llama 3.2 1B Instruct at Inference, $0.01 in and $0.01 out per million tokens, read on 31 Oct 2026.
- Which lab shows up most in this ranking?
- Alibaba Cloud trained 7 of the top 10 models here.
- How is this ranked, and how current is it?
- Every row is the cheapest price for that model with an account behind it: the maker's own, a price below it with a published reason, or a price at or above it. A price below the maker's own with nothing to explain it is listed on the model's page and not ranked. Prices are read every 3 days from 237 providers' public catalogs, per million tokens and blended 3 in : 1 out. Last refresh: 31 Oct 2026.