Read-aloud Assessment Benchmark
Speech models transcribe children reading aloud in Kiswahili. The evaluation measures transcription errors and disagreements with human assessors about which items the child misread.
Fab AI compares 22 model configurations, including Whisper and Gemini, on word error rate and mistake error rate. It published evaluation results in August 2026.
- Made by
- Fab AI
- Released
- August 2026
- Kind
- Speech recognition and translation
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Open-ended
- Measures
- Capability
- Subject
- Education
- Language
- Swahili
- Interaction
- Single turn
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- A host
- In views
- Audio and speech
- Strongest model tested
-
Gemini 3.8 Flash #24 on the capability index
- Last model added
- September 2026
- Scores refreshed
- November 20th, 2026
FastConformer CTC 110M (NeMo) tops the leaderboard with 18.3% Gemini 3.6 Flash leads with 54.5% among models released by September 2026
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is Gemini 3.6 Flash, at 54.5%. This view shows only models we track, each at its best.
The newest model with a score here came out in September 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 1 FastConformer CTC 110M (NeMo) 18.3%
- 2 FastConformer CTC 110M (INT8, Desktop) 23.2%
- 3 FastConformer CTC 110M (INT8, Android) 28.8%
- 4 MAI-Transcribe-2 45.3%
- 5
Gemini 3.6 Flash 54.5%
- 6
Gemini 3.8 Flash 55%
- 7 OmniASR CTC 3B v2 55.9%
- 8
Gemini 3.5 Flash-Lite 66.7%
- 9
Gemini 3.7 Flash 72.4%
- 10 W2V-BERT 2.0 (Swahili) 72.8%
- 11 Sauti ASR 74%
- 12 OmniASR LLM 3B v2 (beam) 75.6%
- 13
Gemma 4 E2B 75.7%
- 14 MMS-1B (Swahili) 75.7%
- 15
Gemini 3.1 Flash-Lite 78%
- 16 Paza Phi-4 Multimodal 88.2%
- 17 Whisper Large (greedy) 174.6%
- 18 Whisper Large v2 (greedy) 221.4%
- 19
Inkling-Small 257.6%
- 20 Whisper Large v2 (beam 5) 399.8%
- 21 Whisper Large (beam 5) 403.8%
- 22 Paza Whisper Large v3 Turbo 1993.5%
Source: Official Read-aloud Assessment Benchmark leaderboard
- 1
Gemini 3.6 Flash 54.5%
- 2
Gemini 3.8 Flash 55%
- 3
Gemini 3.5 Flash-Lite 66.7%
- 4
Gemini 3.7 Flash 72.4%
- 5
Gemma 4 E2B 75.7%
- 6
Gemini 3.1 Flash-Lite 78%
- 7
Inkling-Small 257.6%
Source: Official Read-aloud Assessment Benchmark leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | FastConformer CTC 110M (NeMo) open weights, 0.1B | 18.3% | Official leaderboard | ||
| 2 | FastConformer CTC 110M (INT8, Desktop) open weights, 0.1B | 23.2% | Official leaderboard | ||
| 3 | FastConformer CTC 110M (INT8, Android) open weights, 0.1B | 28.8% | Official leaderboard | ||
| 4 | MAI-Transcribe-2 closed weights | 45.3% | Official leaderboard | ||
| 5 | | 54.5% | Jul 2026 | Official leaderboard | |
| 6 | | 55% | Sep 2026 | Official leaderboard | |
| 7 | OmniASR CTC 3B v2 open weights, 3.1B | 55.9% | Official leaderboard | ||
| 8 | | 66.7% | Jul 2026 | Official leaderboard | |
| 9 | | 72.4% | Aug 2026 | Official leaderboard | |
| 10 | W2V-BERT 2.0 (Swahili) open weights, 0.6B | 72.8% | Official leaderboard | ||
| 11 | Sauti ASR open weights, 0.8B | 74% | Official leaderboard | ||
| 12 | OmniASR LLM 3B v2 (beam) open weights, 4.4B | 75.6% | Official leaderboard | ||
| 13 | | 75.7% | Mar 2026 | Official leaderboard | |
| 14 | MMS-1B (Swahili) open weights, 1B | 75.7% | Official leaderboard | ||
| 15 | | 78% | Mar 2026 | Official leaderboard | |
| 16 | Paza Phi-4 Multimodal open weights, 5.6B | 88.2% | Official leaderboard | ||
| 17 | Whisper Large (greedy) open weights, 1.5B | 174.6% | Official leaderboard | ||
| 18 | Whisper Large v2 (greedy) open weights, 1.5B | 221.4% | Official leaderboard | ||
| 19 | | 257.6% | Jul 2026 | Official leaderboard | |
| 20 | Whisper Large v2 (beam 5) open weights, 1.5B | 399.8% | Official leaderboard | ||
| 21 | Whisper Large (beam 5) open weights, 1.5B | 403.8% | Official leaderboard | ||
| 22 | Paza Whisper Large v3 Turbo open weights, 0.8B | 1993.5% | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 54.5% | Jul 2026 | Official leaderboard | |
| 2 | | 55% | Sep 2026 | Official leaderboard | |
| 3 | | 66.7% | Jul 2026 | Official leaderboard | |
| 4 | | 72.4% | Aug 2026 | Official leaderboard | |
| 5 | | 75.7% | Mar 2026 | Official leaderboard | |
| 6 | | 78% | Mar 2026 | Official leaderboard | |
| 7 | | 257.6% | Jul 2026 | Official leaderboard |
Rankings by part
We read 14 parts from the benchmark’s own leaderboard in 2 groups: wer and mer. Each ranks models on one part of the test. Gemini 3.8 Flash leads 4 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
Passage
7 models we track have a score here.
- 1
Gemini 3.8 Flash ↑1 30.6%
- 2
Gemini 3.6 Flash ↓1 34.2%
- 3
Gemini 3.5 Flash-Lite 37.3%
- 4
Gemini 3.7 Flash 48.9%
- 5
Gemini 3.1 Flash-Lite ↑1 52.9%
- 6
Gemma 4 E2B ↓1 60%
- 7
Inkling-Small 175.4%
Syllables grid
7 models we track have a score here.
- 1
Gemini 3.8 Flash ↑1 52%
- 2
Gemini 3.6 Flash ↓1 55.8%
- 3
Gemma 4 E2B ↑2 77.6%
- 4
Gemini 3.7 Flash 79%
- 5
Gemini 3.5 Flash-Lite ↓2 83.4%
- 6
Gemini 3.1 Flash-Lite 102.1%
- 7
Inkling-Small 230.5%
Syllables isolated
7 models we track have a score here.
- 1
Gemini 3.1 Flash-Lite ↑5 83.1%
- 2
Gemini 3.6 Flash ↓1 88.3%
- 3
Gemini 3.5 Flash-Lite 89.8%
- 4
Gemini 3.8 Flash ↓2 92.5%
- 5
Gemini 3.7 Flash ↓1 93.1%
- 6
Gemma 4 E2B ↓1 103.5%
- 7
Inkling-Small 273.3%
Non-words grid
7 models we track have a score here.
- 1
Gemini 3.6 Flash 64.2%
- 2
Gemini 3.8 Flash 68.2%
- 3
Gemini 3.5 Flash-Lite 70.9%
- 4
Gemma 4 E2B ↑1 83.1%
- 5
Gemini 3.7 Flash ↓1 83.5%
- 6
Gemini 3.1 Flash-Lite 85.6%
- 7
Inkling-Small 252.5%
Non-words isolated
7 models we track have a score here.
- 1
Gemini 3.1 Flash-Lite ↑5 72.3%
- 2
Gemini 3.6 Flash ↓1 79.4%
- 3
Gemini 3.7 Flash ↑1 83.6%
- 4
Gemini 3.5 Flash-Lite ↓1 87.1%
- 5
Gemini 3.8 Flash ↓3 87.1%
- 6
Gemma 4 E2B ↓1 99.5%
- 7
Inkling-Small 220.6%
Letters grid
7 models we track have a score here.
- 1
Gemini 3.6 Flash 48.1%
- 2
Gemini 3.8 Flash 52.9%
- 3
Gemini 3.5 Flash-Lite 67%
- 4
Gemma 4 E2B ↑1 70.7%
- 5
Gemini 3.7 Flash ↓1 74%
- 6
Gemini 3.1 Flash-Lite 76.9%
- 7
Inkling-Small 433.8%
Letters isolated
7 models we track have a score here.
- 1
Gemma 4 E2B ↑4 69.6%
- 2
Gemini 3.1 Flash-Lite ↑4 73.7%
- 3
Gemini 3.6 Flash ↓2 76.9%
- 4
Gemini 3.5 Flash-Lite ↓1 80%
- 5
Gemini 3.8 Flash ↓3 86.7%
- 6
Gemini 3.7 Flash ↓2 90.2%
- 7
Inkling-Small 317.1%
Passage
7 models we track have a score here.
- 1
Gemini 3.8 Flash ↑1 23.9%
- 2
Gemini 3.6 Flash ↓1 26.4%
- 3
Gemini 3.5 Flash-Lite 27.7%
- 4
Gemini 3.7 Flash 43.5%
- 5
Gemma 4 E2B 44.3%
- 6
Gemini 3.1 Flash-Lite 44.5%
- 7
Inkling-Small 150.9%
Syllables grid
7 models we track have a score here.
- 1
Gemini 3.8 Flash ↑1 28.5%
- 2
Gemini 3.6 Flash ↓1 32.8%
- 3
Gemma 4 E2B ↑2 42.7%
- 4
Gemini 3.5 Flash-Lite ↓1 52.2%
- 5
Gemini 3.7 Flash ↓1 58.5%
- 6
Gemini 3.1 Flash-Lite 66.5%
- 7
Inkling-Small 165.1%
Syllables isolated
7 models we track have a score here.
- 1
Gemini 3.1 Flash-Lite ↑5 56.9%
- 2
Gemini 3.6 Flash ↓1 60.5%
- 3
Gemini 3.5 Flash-Lite 60.6%
- 4
Gemini 3.7 Flash 65%
- 5
Gemini 3.8 Flash ↓3 65.2%
- 6
Gemma 4 E2B ↓1 74.6%
- 7
Inkling-Small 240.9%
Non-words grid
7 models we track have a score here.
- 1
Gemini 3.6 Flash 44.6%
- 2
Gemini 3.8 Flash 47.8%
- 3
Gemini 3.5 Flash-Lite 48.5%
- 4
Gemma 4 E2B ↑1 56.9%
- 5
Gemini 3.1 Flash-Lite ↑1 65.3%
- 6
Gemini 3.7 Flash ↓2 69.5%
- 7
Inkling-Small 206.7%
Non-words isolated
7 models we track have a score here.
- 1
Gemini 3.1 Flash-Lite ↑5 46.8%
- 2
Gemini 3.6 Flash ↓1 51.1%
- 3
Gemini 3.5 Flash-Lite 56.2%
- 4
Gemini 3.7 Flash 56.2%
- 5
Gemini 3.8 Flash ↓3 60%
- 6
Gemma 4 E2B ↓1 65.5%
- 7
Inkling-Small 180.1%
Letters grid
7 models we track have a score here.
- 1
Gemini 3.6 Flash 24.2%
- 2
Gemini 3.8 Flash 27.1%
- 3
Gemma 4 E2B ↑2 37.4%
- 4
Gemini 3.5 Flash-Lite ↓1 39%
- 5
Gemini 3.1 Flash-Lite ↑1 45.3%
- 6
Gemini 3.7 Flash ↓2 62.2%
- 7
Inkling-Small 382.4%
Letters isolated
7 models we track have a score here.
- 1
Gemma 4 E2B ↑4 57.5%
- 2
Gemini 3.1 Flash-Lite ↑4 61.6%
- 3
Gemini 3.6 Flash ↓2 65.7%
- 4
Gemini 3.5 Flash-Lite ↓1 68.1%
- 5
Gemini 3.8 Flash ↓3 73.7%
- 6
Gemini 3.7 Flash ↓2 78.9%
- 7
Inkling-Small 301.9%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official Read-aloud Assessment Benchmark leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official Read-aloud Assessment Benchmark leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Audio and video
- Video-MMMU In use 300 expert-level lecture videos with questions testing perception, comprehension and adaptation of knowledge. Accuracy.
Qwen 3.8 Max 88.7%
- LVBench In use Hours-long video understanding with multiple-choice questions on reasoning, grounding and summarization.
Gemini 4 Argon 91.7%
- MMVU In use 3,000 expert-annotated questions on videos across 27 disciplines. Multiple-choice and open-ended accuracy.
Qwen 3.8 Max 82.4%
- MLVU In use Multi-task long video understanding (minutes to hours) with multiple-choice and generation tasks. Accuracy.
Qwen 3.8 Max 90.8%
- FLEURS In use Models transcribe read speech in 102 languages built from FLoRes-101 sentences. Model cards usually report word error rate, so lower scores are better.
Phi-4-multimodal 4.00%
- VoiceBench In use Spoken instructions with speaker, environment and content variation. Voice assistant responses scored by GPT judge or rules.
Qwen3.5-Omni-Plus 93.1%
Data audit
8 of 9 checks pass, 1 warning. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark
Catalog record 1 warning 1 pass
- Warning Task access We don’t know whether the tasks are public.
- Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (22 entries).
- Pass Models scored 7 models we track, plus 15 other entries on its leaderboard.
- Pass Newest score The newest is from September 2026, for Gemini 3.8 Flash.
- Note Frontier models None of the top 10 has a score. The strongest tested is Gemini 3.8 Flash (#24).
- Note Headroom Lower is better, so there’s no ceiling to reach.
Leaderboard import 3 pass
- Pass Importer We read the board on fab-ai.org: 7 rows for models we track and 15 other entries at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.