whatmodel.sh · find the best speech-to-text models you can run

Best speech-to-text models for a 2GB PC without a graphics card

A leaderboard of the best speech-to-text models for a computer with 2GB of RAM and no graphics card. Models are ranked by one score that combines 87 published benchmarks. Models that run on it show the download that fits and the command to run it.

43 models · 87 benchmarks · updated
Filter by your hardware and budget → ~0GB for models (CPU, slow) · reset
ram[2][4][8][16][24][32][48][64][96][128][192][256]
[]
GB
gpu[apple silicon][graphics card][unified memory][none]
price[local][$0.001][$0.002][$0.005][$0.01][any]max per audio min
Filter by task
task[all tasks][long-form transcription][multilingual transcription]
New #1s for your hardware:
orRSS

detecting your OS…

$ terminal

detecting your OS…

Run curl "whatmodel.sh/stt?ram=2&gpu=none&max=any"to get this list in your terminal (add &format=json for agents).

Score vs release date

#modelscore↓where↓RAM↓speed↓get itsources
#modelscore↓where↓RAM↓speed↓get itsources
1Muse Voice Transcribe 1.0 Meta71±3*api $0.0030/audio mindeveloper.meta.com ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)
2Grok Voice Transcribe 2.0 xAI70±3*api $0.0017/audio mindocs.x.ai ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)
4Universal-3.6 Pro Realtime AssemblyAI70±3*api $0.0075/audio minwww.assemblyai.com ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)
5Linden 1 (Agent STT) Speechmatics69±3*api $0.0050/audio minwww.speechmatics.com ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)
6Soniox v5 Async Soniox68±5*api $0.0017/audio minsoniox.com ↗Bangla Speech-to-Text Benchmark v1.0 (Riverborn)
7Ink-2 Cartesia68±3*api $0.0065/audio mindocs.cartesia.ai ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)
8Universal-3.5 Pro AssemblyAI68±3api $0.0035/audio minwww.assemblyai.com ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)verliapp/stt-benchmark · independent STT accuracy benchmarkCoSE-E: A Benchmark for Code-switched Speech Evaluation in Enterprise Settings (arXiv:2609.35645v1), Table 2
9Soniox v5 Real-Time Soniox67±3*api $0.0020/audio minsoniox.com ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)
11Gemini 3.5 Transcribe Google67±6*api $0.0050/audio minai.google.dev ↗Persian ASR Triple Threat Leaderboard (Reza2kn)
12Scribe v2 ElevenLabs66±1api $0.0037/audio minelevenlabs.io ↗Open ASR LeaderboardGigaSpeechBench: A Real-World Multilingual Speech-to-Text BenchmarkVoiceCodeBench (arXiv:2608.28916v2) · structured-token recovery ASR benchmarkCoSE-E: A Benchmark for Code-switched Speech Evaluation in Enterprise Settings (arXiv:2609.35645v1), Table 2BOTNOI Thai telephony ASR report (arXiv:2608.24916v1) · baseline CERsRW-Voice-EQ Bench (Hume AI)+9
13GPT Transcribe OpenAI64±4api $0.0045/audio mindevelopers.openai.com ↗VoiceCodeBench (arXiv:2608.28916v2) · structured-token recovery ASR benchmarkDanish Speech-to-Text Leaderboard (RyeAI) · results dataset
14Zoom Scribe v1 Zoom64±5*api price unknowndevelopers.zoom.us ↗Open ASR Leaderboard
20MAI-Transcribe-1.5 Microsoft AI62±8*api $0.0060/audio minazure.microsoft.com ↗koedesk STT Bench v1 (Track A, FLEURS first-200)
22Fish Audio Transcribe 1 Fish Audio62±5api $0.0060/audio mindocs.fish.audio ↗verliapp/stt-benchmark · independent STT accuracy benchmark
23Inworld STT-1 Inworld62±6*api $0.0025/audio mindocs.inworld.ai ↗verliapp/stt-benchmark · independent STT accuracy benchmark
25Cohere Transcribe (03-2026) Cohere62±2api $0.0042/audio min~5.5GBfal.ai ↗Open ASR Leaderboardtranscribe.cpp model catalog · LibriSpeech test-clean WER of validated full-precision portsDutch ASR Leaderboard (tvosch) · committed result filesOpen Universal Arabic ASR Leaderboard (Elm Research Center)Quranic ASR Leaderboard (Quran-Lab, held-out recitation set)Dysarthric ASR with clinical context on the Speech Accessibility Project (arXiv:2605.02782v2)
27Nova-3 Deepgram61±2api $0.0052/audio mindevelopers.cloudflare.com ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)Russian STT benchmark by domainVoiceCodeBench (arXiv:2608.28916v2) · structured-token recovery ASR benchmarkBangla Speech-to-Text Benchmark v1.0 (Riverborn)verliapp/stt-benchmark · independent STT accuracy benchmarkCoSE-E: A Benchmark for Code-switched Speech Evaluation in Enterprise Settings (arXiv:2609.35645v1), Table 2+6
29Pulse (STT) Smallest.ai61±3*api $0.0040/audio mindocs.smallest.ai ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)
30Chirp 3 Google61±2api price unknowncloud.google.com ↗Open ASR LeaderboardAfriSpeech-MultiBench (arXiv:2511.14255v1) · African-accented English ASR benchmark suiteGigaSpeechBench: A Real-World Multilingual Speech-to-Text BenchmarkVoiceCodeBench (arXiv:2608.28916v2) · structured-token recovery ASR benchmarkBOTNOI Thai telephony ASR report (arXiv:2608.24916v1) · baseline CERsRW-Voice-EQ Bench (Hume AI)+4
32Voxtral Small 24B (2507) Mistral61±2api $0.0040/audio min~59GBdocs.mistral.ai ↗Mistral AI — Voxtral technical reportOpen ASR Leaderboardtranscribe.cpp model catalog · LibriSpeech test-clean WER of validated full-precision portsCoSE-E: A Benchmark for Code-switched Speech Evaluation in Enterprise Settings (arXiv:2609.35645v1), Table 2Danish Speech-to-Text Leaderboard (RyeAI) · results datasetOpen Universal Arabic ASR Leaderboard (Elm Research Center)+1
33Gemini 3.5 Transcribe Live Google61±3*api $0.0090/audio minai.google.dev ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)
39Qwen3-ASR-1.7B Alibaba (Qwen)60±1api $0.0005/audio min~3.3GBdeepinfra.com ↗Qwen/Qwen3-ASR-1.7B model cardOpen ASR Leaderboardtranscribe.cpp model catalog · LibriSpeech test-clean WER of validated full-precision portsGigaSpeechBench: A Real-World Multilingual Speech-to-Text BenchmarkMega-ASR (arXiv:2605.19833v1) · adverse-condition ASR baselines (CHiME-4, VOiCES, NOIZEUS)RW-Voice-EQ Bench (Hume AI)+10
42Speechmatics Enhanced Speechmatics60±2api $0.0067/audio minwww.speechmatics.com ↗Open ASR Leaderboardverliapp/stt-benchmark · independent STT accuracy benchmarkRW-Voice-EQ Bench (Hume AI)Quranic ASR Leaderboard (Quran-Lab, held-out recitation set)
43Deepgram Nova 2 Deepgram60±4api $0.0058/audio mindeepgram.com ↗Russian STT benchmark by domain
46Voxtral Mini Transcribe V2 Mistral59±3api $0.0030/audio mindocs.mistral.ai ↗Voxtral RealtimeDutch ASR Leaderboard (tvosch) · committed result fileskoedesk STT Bench v1 (Track A, FLEURS first-200)
48GPT-Realtime-Whisper OpenAI59±3*api $0.017/audio mindevelopers.openai.com ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)
52Voxtral Mini 3B (2507) Mistral58±1api $0.0010/audio min~12GBdeepinfra.com ↗Mistral AI — Voxtral technical reportOpen ASR LeaderboardAfriSpeech-MultiBench (arXiv:2511.14255v1) · African-accented English ASR benchmark suitetranscribe.cpp model catalog · LibriSpeech test-clean WER of validated full-precision portsMega-ASR (arXiv:2605.19833v1) · adverse-condition ASR baselines (CHiME-4, VOiCES, NOIZEUS)Open ASR Leaderboard – Chilean Spanish (idsudd / Alonso Astroza)+5
53Universal-Streaming (English) AssemblyAI58±3*api $0.0025/audio minwww.assemblyai.com ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)
55Chirp 2 Google58±3api price unknowncloud.google.com ↗Bangla Speech-to-Text Benchmark v1.0 (Riverborn)verliapp/stt-benchmark · independent STT accuracy benchmarkOpen ASR Leaderboard
61Whisper large-v3-turbo OpenAI57±1api $0.0005/audio min~2.4GB~18×developers.cloudflare.com ↗distil-whisper/distil-large-v3.5 model cardOpen ASR Leaderboardtranscribe.cpp model catalog · LibriSpeech test-clean WER of validated full-precision portsverliapp/stt-benchmark · independent STT accuracy benchmarkLiteASR (arXiv:2502.20583v2) · ESB WER of Whisper and LiteASR-compressed encodersRussian ASR Leaderboard (ESpeech; benchmark by Nikolay Shmyrev)+7
64Parakeet TDT 0.6B v3 NVIDIA57±1api $0.0015/audio min~3.5GB~95×www.together.ai ↗NVIDIA ASR model cards (Parakeet / Canary, Hugging Face)Open ASR LeaderboardMoondream parakeet-ultra and parakeet-redux model cards (Benchmarks sections)transcribe.cpp model catalog · LibriSpeech test-clean WER of validated full-precision portsCanary-1B-v2 & Parakeet-TDT-0.6B-v3: Efficient and High-Performance Models for Multilingual ASR and AST (NVIDIA, arXiv:2509.14128v2)Russian STT benchmark by domain+11
66Whisper large-v3 OpenAI57±1api $0.0018/audio min~4.2GBconsole.groq.com ↗Mistral AI — Voxtral technical reportOpen ASR LeaderboardQwen/Qwen3-ASR-1.7B model carddistil-whisper/distil-large-v3.5 model cardAfriSpeech-MultiBench (arXiv:2511.14255v1) · African-accented English ASR benchmark suitetranscribe.cpp model catalog · LibriSpeech test-clean WER of validated full-precision ports+25
75GPT-4o Mini Transcribe OpenAI56±2api $0.0030/audio mindevelopers.openai.com ↗Mistral AI — Voxtral technical reportVoxtral Realtimemu-bench: A Multilingual Utterance Transcription Benchmark (arXiv:2609.32082v1), Table 5Danish Speech-to-Text Leaderboard (RyeAI) · results dataset
82Scribe v2 Realtime ElevenLabs55±3*api $0.0065/audio minelevenlabs.io ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)
89Ink-Whisper Cartesia55±3api price unknowndocs.cartesia.ai ↗Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)verliapp/stt-benchmark · independent STT accuracy benchmark
95GPT-4o Transcribe OpenAI53±1api $0.0060/audio mindevelopers.openai.com ↗Qwen/Qwen3-ASR-1.7B model cardAfriSpeech-MultiBench (arXiv:2511.14255v1) · African-accented English ASR benchmark suitePipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)Qwen-Audio-3.0-ASR Technical Report (arXiv:2609.07549v2)GigaSpeechBench: A Real-World Multilingual Speech-to-Text BenchmarkBangla Speech-to-Text Benchmark v1.0 (Riverborn)+5
98Voxtral Mini 4B Realtime (2602) Mistral53±2api $0.0060/audio min~3.2GBdocs.mistral.ai ↗Open ASR LeaderboardPipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README)Voxtral Realtimetranscribe.cpp model catalog · LibriSpeech test-clean WER of validated full-precision portsVibeVoice-ASR-Streaming Technical Report (arXiv:2609.02812v2), Table 3 single-speaker recognitionDutch ASR Leaderboard (tvosch) · committed result files+1
102Rev AI Machine (Reverb ASR) Rev52±3api $0.0033/audio mindocs.rev.ai ↗Open ASR Leaderboardverliapp/stt-benchmark · independent STT accuracy benchmark
—MAI-Transcribe-2 Microsoft AIno reviewed resultsapi $0.0017/audio minazure.microsoft.com ↗
—Solaria-3 Gladiano reviewed resultsapi $0.010/audio minwww.gladia.io ↗
—Voxtral Mini Transcribe 3 Mistral AIno reviewed resultsapi $0.0035/audio minmistral.ai ↗
—Voxtral Mini Transcribe Realtime 3 Mistral AIno reviewed resultsapi $0.0080/audio minmistral.ai ↗
—Voxtral Small Transcribe 3 Mistral AIno reviewed resultsapi $0.0040/audio minmistral.ai ↗

+ 144 more don't fit your machine or budget · show them

187 of 187 catalogue models listed here · sources and licences

* new or thinly tested: ranked on too few benchmark results · † only the maker’s own numbers so far · ~RAM is estimated; full context may need more memory

43 models · 87 benchmarks · updated