Hopper

The latest open-source models

Hopper is a realtime inference API for open-source models: two LLMs, text-to-speech and speech-to-text, behind a standard chat-completions API. Try the latest open-source models on your own work: in a chat, Hopper runs your prompt or examples on both LLMs side by side and returns each answer with time to first token, tokens per second, token counts and cost, so you can judge the models on your task.

What you can ask

Try the latest open-source LLM on a prompt from my app
Example path → ·

MCP connector

Server URL: https://withhopper.com/mcp
Login: OAuth
Protocol: MCP 2025-11-25
Instructions: Hopper is a realtime inference API for open-source models: DeepSeek V4.1 Flash, including its Oct 16 release (deepseek-v4.1-flash-1017, the newest DeepSeek; deepseek-v4.1-flash is the Sep 10 release), and Gemma 4 31B (gemma-4-31b) for text, Qwen 3 TTS (qwen3-tts) for speech and Nemotron ASR (nemotron-asr) for transcription, behind an OpenAI-compatible API at https://api.withhopper.com/v1. Use generate to run a prompt on one or more of these models, for example to try a newly released model on the user's own prompts or compare models side by side; it returns each answer with time to first token, tokens per second and cost. speak reads text aloud (audio player plus a playable link) and transcribe turns recordings into text with word timestamps. These work right away without an account, within limits; once the user connects their Hopper account, usage is billed to it. When the user wants to use a model from their own app, script or agent, call (it asks them to connect their Hopper account the first time) and give them the base URL, the model id and the snippet from ; the key goes in HOPPER_API_KEY in their environment, never in the chat. lists models and prices; shows credits, usage and keys. No tool deletes, revokes or reveals existing keys or changes account settings: point the user to https://withhopper.com/console/keys for those.

Tools

Reads data4

Changes data4

Skills

4 in the ChatGPT listing

MCP connector

Server URL

https://withhopper.com/mcp

OAuth

Protocol

MCP 2025-11-25

Instructions

Hopper is a realtime inference API for open-source models: DeepSeek V4.1 Flash, including its Oct 16 release (deepseek-v4.1-flash-1017, the newest DeepSeek; deepseek-v4.1-flash is the Sep 10 release), and Gemma 4 31B (gemma-4-31b) for text, Qwen 3 TTS (qwen3-tts) for speech and Nemotron ASR (nemotron-asr) for transcription, behind an OpenAI-compatible API at https://api.withhopper.com/v1. Use generate to run a prompt on one or more of these models, for example to try a newly released model on the user's own prompts or compare models side by side; it returns each answer with time to first token, tokens per second and cost. speak reads text aloud (audio player plus a playable link) and transcribe turns recordings into text with word timestamps. These work right away without an account, within limits; once the user connects their Hopper account, usage is billed to it. When the user wants to use a model from their own app, script or agent, call (it asks them to connect their Hopper account the first time) and give them the base URL, the model id and the snippet from ; the key goes in HOPPER_API_KEY in their environment, never in the chat. lists models and prices; shows credits, usage and keys. No tool deletes, revokes or reveals existing keys or changes account settings: point the user to https://withhopper.com/console/keys for those.

Skills

4 in the ChatGPT listing