Live prices · updated 15 Nov 2026, 10:10 UTC

The same model can cost 1.0× more depending on who serves it.

Enter your daily token volume and see what it costs on every provider for every model, from 124 text models and 22 providers in the public RouterPlus catalog. 53 of those models are sold by more than one provider, and the prices are rarely close.

Biggest gap for this workload

$900/month

GPT-5.6 Sol costs $2,400 on OpenAI (direct API) and $3,300 on Inference.net.

What do you run today?

Where the price gaps are

Monthly cost of your workload on the cheapest and priciest provider of each model, for the 12 models with the biggest price multiple. Hover a row for details.

Llama 3.3 70B InstructMeta · 2 providers
DeepSeek V4 Flash 0731DeepSeek · 9 providers
Qwen3.6 35B A3BQwen · 4 providers
DeepSeek V4.1 Flash 1017DeepSeek · 10 providers
Qwen3.8 27BQwen · 8 providers
GLM 5.3Z.ai · 14 providers
gpt-oss-120bOpenAI · 2 providers
GLM 5.3 FlashZ.ai · 13 providers
Muse Glimmer 30BMeta · 2 providers
Gemma 4 31BGoogle · 5 providers
Kimi K2.6Moonshot AI · 4 providers
GLM 5.2Z.ai · 9 providers
The 12 models with the largest monthly price gap between their cheapest and priciest provider for this workload. The table below lists every model.

Every model, every provider

Monthly cost for your workload. Open a row to compare providers and mark the one you use.

53 models · click a row to see every provider

How the numbers are worked out

Monthly cost = 30 × (uncached input × input price + cached input × cached price + output × output price) ÷ 1M, using each provider’s listed USD per million tokens. When a provider lists no cached price, cached input is billed at the full input price. Only text models billed per token are included; image and video models are priced per image or second.

Source and disclaimer

Prices come from the public RouterPlus model catalog and refresh every 24 hours. This is an independent, unofficial tool, not affiliated with or endorsed by RouterPlus or any provider listed. Prices can change; check the provider before you commit.