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

All results

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	threads	Metrics	Submitted	Details
nemotron-3-5-concurrency	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	librispeech-en	1 *	2	ram_mb: 2523.28 MB cpu_pct: 99.63 %throughput: 122.42 audio-s/s temperature_c: 69.50 °C real_time_factor: 0.0088 ratio	2026-08-21T14:33:18Z	details Profile nemotron-3-5-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 concurrency 1 threads 2 (default 4)
nemotron-3-5-concurrency	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	librispeech-en	2	2	ram_mb: 3651.47 MB cpu_pct: 139.91 %throughput: 142.59 audio-s/s temperature_c: 66.50 °C real_time_factor: 0.0153 ratio	2026-08-21T14:34:39Z	details Profile nemotron-3-5-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 concurrency 2 (default 1) threads 2 (default 4)
nemotron-3-5-concurrency	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	librispeech-en	4	2	ram_mb: 3418.67 MB cpu_pct: 140.64 %throughput: 119.02 audio-s/s temperature_c: 69 °C real_time_factor: 0.0367 ratio	2026-08-21T14:36:13Z	details Profile nemotron-3-5-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 concurrency 4 (default 1) threads 2 (default 4)
nemotron-3-5-concurrency	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	librispeech-en	8	2	ram_mb: 3902.57 MB cpu_pct: 104.18 %throughput: 89.23 audio-s/s temperature_c: 67 °C real_time_factor: 0.0984 ratio	2026-08-21T14:38:20Z	details Profile nemotron-3-5-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 concurrency 8 (default 1) threads 2 (default 4)
nemotron-3-5-de-streaming	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-de	—	2	wer: 0.0520 ratio ram_mb: 2028.16 MB cpu_pct: 98.32 %wer_deletions: 1 count wer_insertions: 3 count real_time_factor: 0.0419 ratio update_frequency: 2.30 Hz partial_stability: 1 ratio wer_substitutions: 9 count first_final_latency: 515.10 ms end_of_speech_latency: 1918.98 ms first_partial_latency: 515.10 ms streaming_responsiveness: 3.94 index	2026-08-21T14:28:38Z	details Profile nemotron-3-5-de-streaming@1.0.0 Pack fleurs-de@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 streaming_latency_ms 320 threads 2 (default 4)
nemotron-3-5-en-streaming	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	librispeech-en	—	2	wer: 0.0687 ratio ram_mb: 2486.57 MB cpu_pct: 93.97 %wer_deletions: 5 count wer_insertions: 0 count real_time_factor: 0.0437 ratio update_frequency: 2.27 Hz partial_stability: 1 ratio wer_substitutions: 18 count first_final_latency: 530.68 ms end_of_speech_latency: 263.83 ms first_partial_latency: 530.68 ms streaming_responsiveness: 3.88 index	2026-08-21T14:29:04Z	details Profile nemotron-3-5-en-streaming@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 streaming_latency_ms 320 threads 2 (default 4)
nemotron-3-5-es-streaming	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-es	—	2	wer: 0.0342 ratio ram_mb: 1989.49 MB cpu_pct: 100.14 %wer_deletions: 1 count wer_insertions: 0 count real_time_factor: 0.0450 ratio update_frequency: 2.29 Hz partial_stability: 1 ratio wer_substitutions: 11 count first_final_latency: 573.95 ms end_of_speech_latency: 915.93 ms first_partial_latency: 573.95 ms streaming_responsiveness: 3.93 index	2026-08-21T14:29:32Z	details Profile nemotron-3-5-es-streaming@1.0.0 Pack fleurs-es@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 streaming_latency_ms 320 threads 2 (default 4)
nemotron-3-5-fr-streaming	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-fr	—	2	wer: 0.0235 ratio ram_mb: 2412.74 MB cpu_pct: 96.82 %wer_deletions: 0 count wer_insertions: 1 count real_time_factor: 0.0430 ratio update_frequency: 2.60 Hz partial_stability: 1 ratio wer_substitutions: 7 count first_final_latency: 461.22 ms end_of_speech_latency: 1076.15 ms first_partial_latency: 461.22 ms streaming_responsiveness: 5.16 index	2026-08-21T14:29:57Z	details Profile nemotron-3-5-fr-streaming@1.0.0 Pack fleurs-fr@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 streaming_latency_ms 320 threads 2 (default 4)
nemotron-3-5-nl-streaming	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-nl	—	2	wer: 0.1314 ratio ram_mb: 2458.34 MB cpu_pct: 103.62 %wer_deletions: 7 count wer_insertions: 3 count real_time_factor: 0.0455 ratio update_frequency: 2.44 Hz partial_stability: 1 ratio wer_substitutions: 26 count first_final_latency: 653.67 ms end_of_speech_latency: 378.63 ms first_partial_latency: 653.67 ms streaming_responsiveness: 4.18 index	2026-08-21T14:30:20Z	details Profile nemotron-3-5-nl-streaming@1.0.0 Pack fleurs-nl@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 streaming_latency_ms 320 threads 2 (default 4)
nemotron-3-5-pt-streaming	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-pt	—	2	wer: 0.0137 ratio ram_mb: 2065.32 MB cpu_pct: 97.45 %wer_deletions: 0 count wer_insertions: 0 count real_time_factor: 0.0421 ratio update_frequency: 2.00 Hz partial_stability: 1 ratio wer_substitutions: 5 count first_final_latency: 530.45 ms end_of_speech_latency: 847.22 ms first_partial_latency: 530.45 ms streaming_responsiveness: 3.44 index	2026-08-21T14:30:48Z	details Profile nemotron-3-5-pt-streaming@1.0.0 Pack fleurs-pt@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 streaming_latency_ms 320 threads 2 (default 4)
nemotron-3-5-de-batch	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-de	—	2	cer: 0.0151 ratio wer: 0.0520 ratio ram_mb: 2964.92 MB cpu_pct: 69.83 %cer_deletions: 10 count temperature_c: 64.50 °C wer_deletions: 1 count cer_insertions: 5 count wer_insertions: 3 count real_time_factor: 0.0073 ratio cer_substitutions: 10 count wer_substitutions: 9 count	2026-08-21T14:26:29Z	details Profile nemotron-3-5-de-batch@1.0.0 Pack fleurs-de@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 threads 2 (default 4)
nemotron-3-5-en-batch	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	librispeech-en	—	2	cer: 0.0206 ratio wer: 0.0687 ratio ram_mb: 3054.59 MB cpu_pct: 66.85 %cer_deletions: 20 count temperature_c: 65 °C wer_deletions: 5 count cer_insertions: 3 count wer_insertions: 0 count real_time_factor: 0.0081 ratio cer_substitutions: 15 count wer_substitutions: 18 count	2026-08-21T14:26:44Z	details Profile nemotron-3-5-en-batch@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 threads 2 (default 4)
nemotron-3-5-es-batch	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-es	—	2	cer: 0.0099 ratio wer: 0.0342 ratio ram_mb: 3504.22 MB cpu_pct: 64.74 %cer_deletions: 7 count temperature_c: 65.50 °C wer_deletions: 1 count cer_insertions: 7 count wer_insertions: 0 count real_time_factor: 0.0077 ratio cer_substitutions: 7 count wer_substitutions: 11 count	2026-08-21T14:27:00Z	details Profile nemotron-3-5-es-batch@1.0.0 Pack fleurs-es@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 threads 2 (default 4)
nemotron-3-5-fr-batch	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-fr	—	2	cer: 0.0100 ratio wer: 0.0235 ratio ram_mb: 2581.94 MB cpu_pct: 71.24 %cer_deletions: 3 count temperature_c: 63 °C wer_deletions: 0 count cer_insertions: 10 count wer_insertions: 1 count real_time_factor: 0.0088 ratio cer_substitutions: 9 count wer_substitutions: 7 count	2026-08-21T14:27:15Z	details Profile nemotron-3-5-fr-batch@1.0.0 Pack fleurs-fr@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 threads 2 (default 4)
nemotron-3-5-nl-batch	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-nl	—	2	cer: 0.0496 ratio wer: 0.1314 ratio ram_mb: 3045.09 MB cpu_pct: 68.46 %cer_deletions: 41 count temperature_c: 65 °C wer_deletions: 7 count cer_insertions: 23 count wer_insertions: 3 count real_time_factor: 0.0090 ratio cer_substitutions: 25 count wer_substitutions: 26 count	2026-08-21T14:27:30Z	details Profile nemotron-3-5-nl-batch@1.0.0 Pack fleurs-nl@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 threads 2 (default 4)
nemotron-3-5-pt-batch	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-pt	—	2	cer: 0.0032 ratio wer: 0.0137 ratio ram_mb: 2914.76 MB cpu_pct: 68.95 %cer_deletions: 1 count temperature_c: 67.50 °C wer_deletions: 0 count cer_insertions: 3 count wer_insertions: 0 count real_time_factor: 0.0079 ratio cer_substitutions: 3 count wer_substitutions: 5 count	2026-08-21T14:27:45Z	details Profile nemotron-3-5-pt-batch@1.0.0 Pack fleurs-pt@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 threads 2 (default 4)
nemotron-3-5-concurrency	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	librispeech-en	1 *	8	ram_mb: 2328.52 MB cpu_pct: 138.94 %throughput: 120.68 audio-s/s temperature_c: 70 °C real_time_factor: 0.0090 ratio	2026-08-21T14:15:32Z	details Profile nemotron-3-5-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 concurrency 1 threads 8 (default 4)
nemotron-3-5-concurrency	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	librispeech-en	2	8	ram_mb: 3023.70 MB cpu_pct: 253.96 %throughput: 133.92 audio-s/s temperature_c: 65.50 °C real_time_factor: 0.0168 ratio	2026-08-21T14:16:52Z	details Profile nemotron-3-5-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 concurrency 2 (default 1) threads 8 (default 4)
nemotron-3-5-concurrency	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	librispeech-en	4	8	ram_mb: 3439.92 MB cpu_pct: 141.21 %throughput: 120.98 audio-s/s temperature_c: 66 °C real_time_factor: 0.0361 ratio	2026-08-21T14:18:28Z	details Profile nemotron-3-5-concurrency@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 concurrency 4 (default 1) threads 8 (default 4)
nemotron-3-5-de-batch	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-de	—	8	cer: 0.0151 ratio wer: 0.0520 ratio ram_mb: 2617.34 MB cpu_pct: 95.96 %cer_deletions: 10 count temperature_c: 67 °C wer_deletions: 1 count cer_insertions: 5 count wer_insertions: 3 count real_time_factor: 0.0073 ratio cer_substitutions: 10 count wer_substitutions: 9 count	2026-08-21T13:17:29Z	details Profile nemotron-3-5-de-batch@1.0.0 Pack fleurs-de@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 threads 8 (default 4)
nemotron-3-5-en-batch	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	librispeech-en	—	8	cer: 0.0206 ratio wer: 0.0687 ratio ram_mb: 2563.90 MB cpu_pct: 102.37 %cer_deletions: 20 count temperature_c: 67 °C wer_deletions: 5 count cer_insertions: 3 count wer_insertions: 0 count real_time_factor: 0.0094 ratio cer_substitutions: 15 count wer_substitutions: 18 count	2026-08-21T13:17:44Z	details Profile nemotron-3-5-en-batch@1.0.0 Pack librispeech-en@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 threads 8 (default 4)
nemotron-3-5-es-batch	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-es	—	8	cer: 0.0099 ratio wer: 0.0342 ratio ram_mb: 3634.93 MB cpu_pct: 88.84 %cer_deletions: 7 count temperature_c: 66 °C wer_deletions: 1 count cer_insertions: 7 count wer_insertions: 0 count real_time_factor: 0.0087 ratio cer_substitutions: 7 count wer_substitutions: 11 count	2026-08-21T13:18:00Z	details Profile nemotron-3-5-es-batch@1.0.0 Pack fleurs-es@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 threads 8 (default 4)
nemotron-3-5-fr-batch	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-fr	—	8	cer: 0.0100 ratio wer: 0.0235 ratio ram_mb: 2206.50 MB cpu_pct: 80.50 %cer_deletions: 3 count temperature_c: 67 °C wer_deletions: 0 count cer_insertions: 10 count wer_insertions: 1 count real_time_factor: 0.0092 ratio cer_substitutions: 9 count wer_substitutions: 7 count	2026-08-21T13:18:15Z	details Profile nemotron-3-5-fr-batch@1.0.0 Pack fleurs-fr@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 threads 8 (default 4)
nemotron-3-5-nl-batch	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-nl	—	8	cer: 0.0496 ratio wer: 0.1314 ratio ram_mb: 3448.91 MB cpu_pct: 87.59 %cer_deletions: 41 count temperature_c: 64.42 °C wer_deletions: 7 count cer_insertions: 23 count wer_insertions: 3 count real_time_factor: 0.0101 ratio cer_substitutions: 25 count wer_substitutions: 26 count	2026-08-21T13:18:31Z	details Profile nemotron-3-5-nl-batch@1.0.0 Pack fleurs-nl@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 threads 8 (default 4)
nemotron-3-5-pt-batch	nemotron / nemotron-3.5-asr-streaming-0.6bcuda	nvidia-rtx-5090	fleurs-pt	—	8	cer: 0.0032 ratio wer: 0.0137 ratio ram_mb: 2563.79 MB cpu_pct: 100.57 %cer_deletions: 1 count temperature_c: 66.50 °C wer_deletions: 0 count cer_insertions: 3 count wer_insertions: 0 count real_time_factor: 0.0085 ratio cer_substitutions: 3 count wer_substitutions: 5 count	2026-08-21T13:18:46Z	details Profile nemotron-3-5-pt-batch@1.0.0 Pack fleurs-pt@1.0.0 Repeats 2 Backend cuda Submitted by taktx#fbc9efc5 threads 8 (default 4)

Constant across this view: streaming_latency_ms 320

* profile default

1–25 of 934

← PrevPage 1 of 38Next →