Models & Providers

OpenWhispr supports multiple speech-to-text engines and AI providers. Run everything locally for privacy, use cloud APIs for speed, or mix both.

🖥️Local Transcription Models

Run speech-to-text entirely on your machine. No internet required, no data leaves your device.

OpenAI Whisper

via whisper.cpp · 99+ languages · GGML format

Model	Size	Speed	Quality	Best for
Tiny	75 MB	Fastest	Basic	Lightweight model for quick drafts and low-powered hardware
BaseRecommended	142 MB	Fast	Good	Best balance of speed and accuracy for most users
Small	466 MB	Medium	Better	Improved accuracy for professional transcription
Medium	1.5 GB	Slow	High	High accuracy for multilingual and technical content
Large v3	3 GB	Slowest	Best	Maximum accuracy across all 99+ languages
Turbo	1.6 GB	Fast	Good	Large-v3 architecture optimized for speed — near-large accuracy at base-level latency

NVIDIA Parakeet

via sherpa-onnx · 25 languages · ONNX INT8

Parakeet TDT 0.6B v3RecommendedState-of-the-art English

State-of-the-art English accuracy with 25-language multilingual support. INT8 quantized for efficient local inference.

680 MB Fast 25 languages 100% local

Parakeet languages: English, Bulgarian, Croatian, Czech, Danish, Dutch, Estonian, Finnish, French, German, Greek, Hungarian, Italian, Latvian, Lithuanian, Maltese, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Spanish, Swedish, Ukrainian.

☁️Cloud Transcription Providers

Use your own API keys for cloud-powered transcription. Bring-your-own-key keeps you in control.

OpenAI

GPT-4o Mini Transcribe

Fast, accurate transcription

Very Fast

GPT-4o Transcribe

Most accurate OpenAI transcription

Standard

Whisper

Original Whisper API endpoint

Standard

Get API key

Groq

Fastest

Whisper Large v3 Turbo

216x real-time speed — the fastest cloud transcription available

Ultra-Fast

Get API key

Mistral

Voxtral Mini

Multilingual transcription from Mistral AI

Fast

Get API key

Custom Endpoint

Any OpenAI-compatible API

Connect Ollama, self-hosted Whisper servers, LocalAI, or any service with an OpenAI-compatible /audio/transcriptions endpoint.

✨OpenWhispr Cloud

Zero-config transcription and AI processing. No API keys needed — just sign in and go.

Managed transcription

We route to the fastest provider automatically

AI text cleanup included

Formatting, punctuation, and filler removal

Free tier available

Upgrade to Pro for unlimited usage

🧠AI Text Processing

After transcription, OpenWhispr can clean up your text — fixing grammar, removing filler words, and formatting output. Choose from cloud or local AI models.

Cloud Providers (BYOK)

OpenAI

GPT-5.2, GPT-5 Mini, GPT-5 Nano, GPT-4.1, GPT-4.1 Mini, GPT-4.1 Nano

Anthropic

Claude Opus 4.5, Claude Sonnet 4.5, Claude Haiku 4.5

Google

Gemini 3 Pro, Gemini 3 Flash, Gemini 2.5 Flash Lite

Groq

Qwen3 32B, GPT-OSS 120B, GPT-OSS 20B, LLaMA 3.3 70B, LLaMA 3.1 8B, Mixtral 8x7B

Local Models (via llama.cpp)

Qwen

Qwen3 8B

0.6B, 1.7B, 4B, 8B, 32B variants

Mistral

Mistral 7B Instruct

Q4 and Q5 quantizations

Meta LLaMA

LLaMA 3.2 3B

1B, 3B, 8B variants

OpenAI OSS

GPT-OSS 20B

Open-source flagship

Local reasoning models run via llama.cpp with GGUF quantization. All processing stays on your machine.

⚙️How It Fits Together

1. You speak

Hold your hotkey and talk naturally into any app

2. Model transcribes

Whisper, Parakeet, or a cloud provider converts speech to text

3. AI cleans up

Optional AI processing fixes grammar, removes filler, and formats your text

Try every model for free

Download OpenWhispr and choose the models that work best for you. Local models are completely free. Cloud providers use your own API keys.

Loading... GitHub

GitHub