# Jais-ASR Benchmark Scores & Performance

> Jais-ASR has IFM Arabic ASR evaluation results. Voice evidence is display-only and does not produce a text-model score.

Canonical page: https://benchlm.ai/models/jais-asr

Last updated: October 15, 2026

## Model Details

| Property | Value |
|----------|-------|
| Creator | IFM |
| Source Type | Open Weight |
| Reasoning Type | Non-Reasoning |
| Context Window | N/A |
| Official model card | [IFM model documentation](https://huggingface.co/IFM/Jais-ASR) |
| Overall Score | Not scored (voice evidence only) |
| Overall Rank | Unranked |

## Family & Coverage

- Family: Jais-ASR
- Variant: base
- Benchmarks covered: 0 of 638
- Coverage note: Speech and voice results appear below; no weighted text-model benchmark rows are stored.

## Speech and voice evidence

[Jais-ASR Arabic speech evaluation](/voice-benchmarks/jais-asr-arabic-evaluation)

IFM evaluated Jais-ASR using the public leaderboard script. These are provider-reported measurements, not independently verified leaderboard submissions. The six-set Arabic macro-average, Mixat, and internal Arabic Speech Benchmark have different samples and aggregation rules and are not pooled. Speech error rates stay separate from weighted text-model scores. October 15 is the source retrieval date, not an established evaluation or first-release day.

| Metric | Result |
| --- | --- |
| Six-set Arabic macro-average WER | 22.85% |
| Six-set Arabic macro-average CER | 8.37% |
| Mixat code-switched WER | 17.8% |
| Mixat code-switched CER | 6.0% |

IFM evaluation retrieved October 15, 2026. Lower error rates are better. Six-set Arabic macro-averages and Mixat code-switched results are separate protocols; neither enters general model rankings.

- [Evaluation source](https://ifm.ai/blog/jais-asr/)

Apache 2.0 speech recognition weights built on a Jais-3 language backbone, with 11.4 GB BF16 weights. The model supports Modern Standard Arabic and dialects across 19 Arab countries, plus English, French, Spanish, and Italian. IFM supplies Transformers examples and a vLLM transcription plugin. The launch blog establishes October 2026, without an exact day.

## Other IFM Models

- [Jais-3-MoVA-36B-A4B](/models/jais-3-mova-36b-a4b) - Score: not computed
- [Jais-3-32B](/models/jais-3-32b) - Score: not computed
- [Jais-3-375B-A23B](/models/jais-3-375b-a23b) - Score: not computed
- [Jais-3-7B](/models/jais-3-7b) - Score: not computed
- [Jais-3-3.7B](/models/jais-3-3-7b) - Score: not computed
- [Jais-3-0.9B](/models/jais-3-0-9b) - Score: not computed
- [Jais-3-0.9B-Uno](/models/jais-3-0-9b-uno) - Score: not computed
- [Jais-3-3.7B-Uno](/models/jais-3-3-7b-uno) - Score: not computed
- [Jais-3-32B-Uno](/models/jais-3-32b-uno) - Score: not computed
- [Jais-3-7B-Uno](/models/jais-3-7b-uno) - Score: not computed