GOESB — Global Open Edge Speech Benchmark · pronounced “goose-bee”

Leaderboard

Objective, reproducible benchmarks for local (edge) speech AI. Every result here re-verified its hash and signature before being accepted — see why you can trust the numbers. New here? Read the docs.

This is every verified result, filterable and sortable. Looking for a single best recommendation for your constraints instead? Try the homepage →

Lowest energy Fastest (lowest RTF)Most accurate (lowest WER)

AnyEvery verified result, batch and streaming mixed together.

Any Batch Streaming Concurrency

Backend: cuda

LanguagePackRuntimeModelBackendHardware

Clear + Add filter

Quality gates:WER < 5%WER < 10%WER < 15%

custom…

Sorted by most recently submitted · 25 results · click a metric below to sort by it

Profile	Runtime / Model	Hardware	Pack	concurrency	Metrics	Submitted	Details
whisper-tiny-en-batch	faster-whisper / whisper-tinycuda	nvidia-rtx-5090	librispeech-en	—	cer: 0.0514 ratio wer: 0.1104 ratio ram_mb: 865.29 MB cpu_pct: 585.97 %temperature_c: 53.25 °C real_time_factor: 0.0205 ratio	2026-08-01T14:02:28Z	details Profile whisper-tiny-en-batch@1.1.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 quantization int8 threads 4 vad true
whisper-tiny-es-batch	faster-whisper / whisper-tinycuda	nvidia-rtx-5090	fleurs-es	—	cer: 0.0263 ratio wer: 0.0997 ratio ram_mb: 798.54 MB cpu_pct: 543.39 %temperature_c: 53 °C real_time_factor: 0.0187 ratio	2026-08-01T14:02:37Z	details Profile whisper-tiny-es-batch@1.1.0 Pack fleurs-es@1.0.0 Repeats 2 Backend cuda beam_size 5 quantization int8 threads 4 vad true
whisper-tiny-fr-batch	faster-whisper / whisper-tinycuda	nvidia-rtx-5090	fleurs-fr	—	cer: 0.1334 ratio wer: 0.4047 ratio ram_mb: 829.09 MB cpu_pct: 516.24 %temperature_c: 55 °C real_time_factor: 0.0205 ratio	2026-08-01T14:02:45Z	details Profile whisper-tiny-fr-batch@1.1.0 Pack fleurs-fr@1.0.0 Repeats 2 Backend cuda beam_size 5 quantization int8 threads 4 vad true
whisper-tiny-nl-batch	faster-whisper / whisper-tinycuda	nvidia-rtx-5090	fleurs-nl	—	cer: 0.1955 ratio wer: 0.4708 ratio ram_mb: 830.74 MB cpu_pct: 531.77 %temperature_c: 55 °C real_time_factor: 0.0252 ratio	2026-08-01T14:02:53Z	details Profile whisper-tiny-nl-batch@1.1.0 Pack fleurs-nl@1.0.0 Repeats 2 Backend cuda beam_size 5 quantization int8 threads 4 vad true
whisper-tiny-pt-batch	faster-whisper / whisper-tinycuda	nvidia-rtx-5090	fleurs-pt	—	cer: 0.0309 ratio wer: 0.1230 ratio ram_mb: 798.13 MB cpu_pct: 505.11 %temperature_c: 53.19 °C real_time_factor: 0.0176 ratio	2026-08-01T14:03:01Z	details Profile whisper-tiny-pt-batch@1.1.0 Pack fleurs-pt@1.0.0 Repeats 2 Backend cuda beam_size 5 quantization int8 threads 4 vad true
whisper-base-concurrency	faster-whisper / whisper-basecuda	nvidia-rtx-5090	librispeech-en	1 *	ram_mb: 920.80 MB cpu_pct: 602.63 %throughput: 46.83 audio-s/s temperature_c: 53.38 °C real_time_factor: 0.0249 ratio	2026-08-01T14:52:20Z	details Profile whisper-base-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 1 quantization int8 threads 4 vad true
whisper-base-concurrency	faster-whisper / whisper-basecuda	nvidia-rtx-5090	librispeech-en	2	ram_mb: 962.81 MB cpu_pct: 621.08 %throughput: 64.69 audio-s/s temperature_c: 53.75 °C real_time_factor: 0.0372 ratio	2026-08-01T14:53:22Z	details Profile whisper-base-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 2 (default 1) quantization int8 threads 4 vad true
whisper-base-concurrency	faster-whisper / whisper-basecuda	nvidia-rtx-5090	librispeech-en	4	ram_mb: 1033.38 MB cpu_pct: 644.57 %throughput: 73.77 audio-s/s temperature_c: 54.63 °C real_time_factor: 0.0670 ratio	2026-08-01T14:54:25Z	details Profile whisper-base-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 4 (default 1) quantization int8 threads 4 vad true
whisper-base-concurrency	faster-whisper / whisper-basecuda	nvidia-rtx-5090	librispeech-en	8	ram_mb: 1208.91 MB cpu_pct: 674.50 %throughput: 77.50 audio-s/s temperature_c: 53.63 °C real_time_factor: 0.1255 ratio	2026-08-01T14:55:29Z	details Profile whisper-base-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 8 (default 1) quantization int8 threads 4 vad true
whisper-large-v3-concurrency	faster-whisper / whisper-large-v3cuda	nvidia-rtx-5090	librispeech-en	1 *	ram_mb: 3549.51 MB cpu_pct: 387.16 %throughput: 28.64 audio-s/s temperature_c: 54.38 °C real_time_factor: 0.0423 ratio	2026-08-01T14:56:38Z	details Profile whisper-large-v3-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 1 quantization int8 threads 4 vad true
whisper-large-v3-concurrency	faster-whisper / whisper-large-v3cuda	nvidia-rtx-5090	librispeech-en	2	ram_mb: 3593.76 MB cpu_pct: 443.39 %throughput: 33.60 audio-s/s temperature_c: 54.81 °C real_time_factor: 0.0705 ratio	2026-08-01T14:57:49Z	details Profile whisper-large-v3-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 2 (default 1) quantization int8 threads 4 vad true
whisper-large-v3-concurrency	faster-whisper / whisper-large-v3cuda	nvidia-rtx-5090	librispeech-en	4	ram_mb: 3596.53 MB cpu_pct: 504.04 %throughput: 36.97 audio-s/s temperature_c: 55 °C real_time_factor: 0.1281 ratio	2026-08-01T14:59:00Z	details Profile whisper-large-v3-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 4 (default 1) quantization int8 threads 4 vad true
whisper-large-v3-concurrency	faster-whisper / whisper-large-v3cuda	nvidia-rtx-5090	librispeech-en	8	ram_mb: 3725.68 MB cpu_pct: 499.60 %throughput: 32.32 audio-s/s temperature_c: 54.75 °C real_time_factor: 0.2879 ratio	2026-08-01T15:00:13Z	details Profile whisper-large-v3-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 8 (default 1) quantization int8 threads 4 vad true
whisper-medium-concurrency	faster-whisper / whisper-mediumcuda	nvidia-rtx-5090	librispeech-en	1 *	ram_mb: 2105.89 MB cpu_pct: 433.56 %throughput: 30.88 audio-s/s temperature_c: 54.31 °C real_time_factor: 0.0386 ratio	2026-08-01T15:01:20Z	details Profile whisper-medium-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 1 quantization int8 threads 4 vad true
whisper-medium-concurrency	faster-whisper / whisper-mediumcuda	nvidia-rtx-5090	librispeech-en	2	ram_mb: 2051.79 MB cpu_pct: 466.35 %throughput: 36.90 audio-s/s temperature_c: 55.13 °C real_time_factor: 0.0639 ratio	2026-08-01T15:02:27Z	details Profile whisper-medium-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 2 (default 1) quantization int8 threads 4 vad true
whisper-medium-concurrency	faster-whisper / whisper-mediumcuda	nvidia-rtx-5090	librispeech-en	4	ram_mb: 2158.09 MB cpu_pct: 538.84 %throughput: 40.60 audio-s/s temperature_c: 55.56 °C real_time_factor: 0.1125 ratio	2026-08-01T15:03:34Z	details Profile whisper-medium-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 4 (default 1) quantization int8 threads 4 vad true
whisper-medium-concurrency	faster-whisper / whisper-mediumcuda	nvidia-rtx-5090	librispeech-en	8	ram_mb: 2122.76 MB cpu_pct: 536.60 %throughput: 35.14 audio-s/s temperature_c: 55.69 °C real_time_factor: 0.2586 ratio	2026-08-01T15:04:44Z	details Profile whisper-medium-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 8 (default 1) quantization int8 threads 4 vad true
whisper-small-concurrency	faster-whisper / whisper-smallcuda	nvidia-rtx-5090	librispeech-en	1 *	ram_mb: 1142.67 MB cpu_pct: 528.06 %throughput: 40.31 audio-s/s temperature_c: 54 °C real_time_factor: 0.0292 ratio	2026-08-01T15:05:47Z	details Profile whisper-small-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 1 quantization int8 threads 4 vad true
whisper-small-concurrency	faster-whisper / whisper-smallcuda	nvidia-rtx-5090	librispeech-en	2	ram_mb: 1160.53 MB cpu_pct: 574.03 %throughput: 52.08 audio-s/s temperature_c: 55.19 °C real_time_factor: 0.0457 ratio	2026-08-01T15:06:52Z	details Profile whisper-small-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 2 (default 1) quantization int8 threads 4 vad true
whisper-small-concurrency	faster-whisper / whisper-smallcuda	nvidia-rtx-5090	librispeech-en	4	ram_mb: 1222.39 MB cpu_pct: 594.90 %throughput: 58.60 audio-s/s temperature_c: 54.19 °C real_time_factor: 0.0801 ratio	2026-08-01T15:07:56Z	details Profile whisper-small-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 4 (default 1) quantization int8 threads 4 vad true
whisper-small-concurrency	faster-whisper / whisper-smallcuda	nvidia-rtx-5090	librispeech-en	8	ram_mb: 1303.44 MB cpu_pct: 603.23 %throughput: 55.05 audio-s/s temperature_c: 53.63 °C real_time_factor: 0.1670 ratio	2026-08-01T15:09:02Z	details Profile whisper-small-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 8 (default 1) quantization int8 threads 4 vad true
whisper-tiny-concurrency	faster-whisper / whisper-tinycuda	nvidia-rtx-5090	librispeech-en	1 *	ram_mb: 854.65 MB cpu_pct: 634.52 %throughput: 52.01 audio-s/s temperature_c: 55.75 °C real_time_factor: 0.0225 ratio	2026-08-01T15:10:04Z	details Profile whisper-tiny-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 1 quantization int8 threads 4 vad true
whisper-tiny-concurrency	faster-whisper / whisper-tinycuda	nvidia-rtx-5090	librispeech-en	2	ram_mb: 905.04 MB cpu_pct: 668.18 %throughput: 72.35 audio-s/s temperature_c: 54.81 °C real_time_factor: 0.0338 ratio	2026-08-01T15:11:07Z	details Profile whisper-tiny-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 2 (default 1) quantization int8 threads 4 vad true
whisper-tiny-concurrency	faster-whisper / whisper-tinycuda	nvidia-rtx-5090	librispeech-en	4	ram_mb: 979.58 MB cpu_pct: 656.53 %throughput: 80.08 audio-s/s temperature_c: 56.38 °C real_time_factor: 0.0629 ratio	2026-08-01T15:12:09Z	details Profile whisper-tiny-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 4 (default 1) quantization int8 threads 4 vad true
whisper-tiny-concurrency	faster-whisper / whisper-tinycuda	nvidia-rtx-5090	librispeech-en	8	ram_mb: 1146.46 MB cpu_pct: 681.02 %throughput: 81.82 audio-s/s temperature_c: 57.25 °C real_time_factor: 0.1192 ratio	2026-08-01T15:13:12Z	details Profile whisper-tiny-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda beam_size 5 concurrency 8 (default 1) quantization int8 threads 4 vad true

Constant across this view: beam_size 5 · quantization int8 · threads 4 · vad true

* profile default

26–50 of 130

← PrevPage 2 of 6Next →