Skip to content
Tokenly.

Tokenly

Pricing, stated plainly

A price only appears here when a record carries a verified amount, unit and currency. Everything else is labelled.

How units are quoted

Per million tokens
Most chat models are priced per million input and output tokens. Token counts come from the provider's tokenizer, which can differ between models.
Cache reads
Where a provider offers prompt caching, the cache-read price is listed separately. Absence of a cache price means caching is not confirmed, not that it is free.
Embeddings
Embedding models are priced per million input tokens only. There is no output price to compare.

Browse models

Pricing, stated plainly
ModelDeveloperModalitiesContextInput price
BGE-M3BAAI/bge-m3BAAItext8.2K tokensNot yet published
DeepSeek V3deepseek-chatDeepSeektext · code131K tokensPending verification
Gemma 2 27B ITgoogle/gemma-2-27b-itGoogle DeepMindtext8.2K tokensNot yet published
Llama 3.1 70B Instructmeta-llama/Llama-3.1-70B-InstructMetatext128K tokensNot yet published
Llama 3.2 3B Instructmeta-llama/Llama-3.2-3B-InstructMetatext128K tokensPending verification
Mistral Large 2mistral-large-2407Mistral AItext128K tokensNot yet published
Mixtral 8x7B Instructmistralai/Mixtral-8x7B-Instruct-v0.1Mistral AItext33K tokensNot yet published
Phi-3.5 Mini Instructmicrosoft/Phi-3.5-mini-instructMicrosofttext131K tokensNot yet published
Qwen2.5 72B InstructQwen/Qwen2.5-72B-InstructAlibaba Qwentext131K tokensNot yet published
Qwen2.5 Coder 32B InstructQwen/Qwen2.5-Coder-32B-InstructAlibaba Qwentext · code131K tokensNot yet published

Platform fees

Tokenly aims to route requests through one account. Any platform fee will be published here before billing begins. Phase 1 involves no billing, because there is no request routing yet.

“Not yet published” is a real state

A zero in a price column reads as free. Where a price has not been verified, the catalogue says so, and price sorting places unverified models after verified ones.