Qwen 2 Audio 7B

Aug 2024
Qwen 2 Audio 7B
Compare in
on
self-host onlyprice
80.8MAEB v1 audio-only
Aug 2024released
131K tokensMax input
enLanguages

Ranking?

Full ranking →
1E5 Omni 7B56.12LCO Embedding Omni 7B55.23LCO Embedding Omni 3B54.54EmbeddingGemma 2535E5 Omni 3B51.2
···
23Speecht5 Multimodal7.924Qwen 2 Audio 7B5.3

Alternatives?

ModelScoreCostPriceWhat changes
BetterEmbeddingGemma 2unstatedOpen×10—$0.02Max input: 131K tokens → 8K tokens+1HOpenE5 Omni 7Bunstated×10.6——Max input: 131K tokens → 33K tokens+2BetterJina Embeddings v5 Omni NanounstatedOpen×9.5—$0.02Max input: 131K tokens → 8K tokens+1BetterJina Embeddings v5 Omni SmallunstatedOpen×9.4—$0.05Max input: 131K tokens → 33K tokens+1LBetterLCO Embedding Omni 7BunstatedOpen×10.4——Max input: 131K tokens → 33K tokens+2

Run it yourself

estimate
Fits on✓ 8 GB (phone, laptop) · 8-bit✓ 16 GB · 8-bit✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceMITfree for commercial use

Benchmarks

in Embeddings & Retrieval
MAEB v1 · Mean over tasks · multilingual (159 languages)34.5#13
MAEB v1 · Mean over task types · multilingual (159 languages)37#18
MAEB v1 · Audio Classification · multilingual (102 languages)58.3#2
MAEB v1 · Audio Zero-shot Classification · English12.4#24
MAEB v1 · Any-to-any Retrieval · multilingual (150 languages)5.3#24
MAEB v1 · Audio Pair Classification · English56.9#12
MAEB v1 · Audio Reranking · No linguistic content80.8#9
MAEB v1 · Audio Multilabel Classification · English32.6#24
MAEB v1 · Audio Clustering · multilingual (6 languages)12.7#5
MAEB v1 audio-only · Mean over tasks · multilingual (102 languages)51#3
MAEB v1 audio-only · Mean over task types · multilingual (102 languages)49.3#22
MAEB v1 audio-only · Audio Classification · multilingual (102 languages)58.3#2
MAEB v1 audio-only · Any-to-any Retrieval · multilingual (4 languages)37.6#40
MAEB v1 audio-only · Audio Clustering · multilingual (6 languages)12.7#5
MAEB v1 audio-only · Audio Pair Classification · English56.9#12
MAEB v1 audio-only · Audio Reranking · No linguistic content80.8#9

FAQ

How much does Qwen 2 Audio 7B cost?›

There is no per-unit API price: self-host only.

Is Qwen 2 Audio 7B free to use?›

There is no free API. The weights are open, so you can run it yourself.

How good is Qwen 2 Audio 7B, and what is it best at?›

#8 of 27 in Multimodal.

Is Qwen 2 Audio 7B better than Speecht5 Multimodal?›

On MAEB v1 · Any-to-any Retrieval · multilingual (150 languages), Qwen 2 Audio 7B scores 5.3 vs 7.9 — lower.

When was Qwen 2 Audio 7B released? Is there a newer version?›

Released 2024-08-09. It is the current version in its line.

Is Qwen 2 Audio 7B open source?›

Yes, the weights are open (MIT licence).

Can I run Qwen 2 Audio 7B locally?›

The weights are open and can be self-hosted; hardware needs are not confirmed yet.

What is Qwen 2 Audio 7B?›

Qwen 2 Audio 7B is an open-weights embeddings & retrieval model from Alibaba (Qwen), released Aug 2024.

What are the best alternatives to Qwen 2 Audio 7B?›

EmbeddingGemma 2 (better), E5 Omni 7B (open weights).