小马算力

Editor Review

To move existing code, replace the request address with the platform address in the docs and replace the key with one created in the console. Streaming is a flag on the request. The reasoning toggle sits in an extra field, and turning it off drops the reasoning trace from the answer. The docs also ask for a provider id and say the endpoint handles fallback and cost selection. The key page creates and lists keys. Model names in samples should be replaced with names from the model list. It fits developers who already call a chat-completions API and want to change model source with a small edit.

AI Tools Navigator Editorial TeamUpdated: 2026-10-09

What is 小马算力

TokenPony offers one chat API so an existing app can reach multiple models through a single address. The quick start says it follows a common chat-completions format: change the request address and the key to the values the platform assigns. A call names the model, can stream, can toggle a reasoning trace in an extra field, and includes a provider id. The docs say the single endpoint handles fallback and picks a lower-cost option. You register first, then create and store a key. The quick start does not list prices.

小马算力 Key Features

Chat-completions compatible calls

Existing code starts a chat after the request address and key are changed to the platform address and a key from the console.

Streaming switch

The request can turn streaming on or off. The quick start shows a non-streaming sample and how to switch it to streaming.

Reasoning toggle and provider id

Extra fields can turn the reasoning trace on or off and set a provider id for that request.

Create and store API keys

After sign-in, create a key in the key menu. The docs say to keep it in an environment variable and out of public code.

小马算力 Key Advantages

Existing chat-completions code only needs a new address and key, and the docs include Python, Node, and command-line samples

A request can switch streaming and use extra fields for the reasoning trace and provider id

The docs say the single endpoint handles fallback and selects a lower-cost option

Keys are created and viewed on their own page, and the docs say not to put them in public code

小马算力 Use Cases

Point an existing chat client here

A developer keeps the existing call shape and changes only the address, key, and model name so the app uses TokenPony.

Toggle reasoning per request

The same code turns the reasoning trace on when that process should be visible, and off when only the final answer is needed.

Manage call keys in one place

The team creates, views, and retires keys on the key page so they are not scattered through project source.

Frequently Asked Questions

▶What has to change in chat code that already exists?

The quick start names two required edits: set the request address to the platform address in the docs, and set the key to one created at TokenPony. Change the model name to the one you want, and turn on streaming if needed. The reasoning toggle and provider id go in extra fields. Keep the new key in an environment variable and do not commit it.

▶Where are the reasoning trace and streaming set?

Streaming is a flag on the request. The sample leaves it off, and turning it on returns a stream. The reasoning trace is a separate switch in the extra fields, and the sample shows how to turn it off. The two are independent: you can stream only the final answer, or keep the reasoning trace on a non-streaming response. Field names follow the quick start sample.

Editor Review

What is 小马算力

Basic Info

Best For

小马算力 Key Features

小马算力 Key Advantages

小马算力 Use Cases

Frequently Asked Questions

Latest News

User Reviews

Coze

SiliconFlow

码上飞

Zion

OpenRouter

秒哒

Dify

Trickle AI

小马算力

Editor Review

What is 小马算力

Basic Info

Best For

小马算力 Key Features

小马算力 Key Advantages

小马算力 Use Cases

Frequently Asked Questions

Latest News

User Reviews

Related Tools

Coze

SiliconFlow

码上飞

Zion

OpenRouter

秒哒

Dify

Trickle AI