What does AudioMC (audio output) measure?

The AudioMC questions with the model required to answer in spoken audio. Scores are reported in percent of tasks solved; higher is better.

Which AI model leads AudioMC (audio output)?

GPT Realtime 2 leads AudioMC (audio output) with 48.5% as of 19 Sept 2026, ahead of Tml Interaction Small at 43.4%.

How many models have AudioMC (audio output) results?

15 model configurations have a AudioMC (audio output) result on BenchLeader, all taken from Scale AI SEAL.

Who runs AudioMC (audio output) and how often is it updated?

AudioMC (audio output) is published by Scale AI SEAL. BenchLeader re-reads the published results every morning and records the date each result was published.

Does AudioMC (audio output) count toward the BenchLeader Index?

No. AudioMC (audio output) is shown for reference but left out of the composite index.

AudioMC (audio output)

AudioMC answered in speech. Scale AI.

As of 19 Sept 2026, GPT Realtime 2 leads AudioMC (audio output) on BenchLeader with 48.5%, ahead of Tml Interaction Small at 43.4%, across 15 model configurations with a published result.

Published by: Scale AI SEAL
Category: Multimodal
Index weight: Reference only
Models: 15
Data as of: 19 Sept 2026

Scale AI SEAL Leaderboards.

What the test looks like

The AudioMC questions with the model required to answer in spoken audio.

How it is scored

Accuracy of the spoken answer, published by Scale AI.

What to keep in mind

Speech-to-speech models only.

Top 15

1GPT Realtime 2 (xhigh)48.5%
2Tml Interaction Small43.4%
3GPT Realtime 237.6%
4Gemini 3.1 Flash Live (thinking)36.1%
5GPT Realtime 1.534.7%
6Gemini 3.1 Flash Live26.8%
7Qwen3 Omni 30B A3B Instruct24.3%
8GPT 4o Audio23.2%
9Gemini 2.5 Flash Native Audio 12 2025 (thinking)21.5%
10GPT Realtime20.4%
11GPT Realtime Mini13.9%
12Gemini 2.5 Flash Native Audio 12 2025 (none)13.9%
13GPT 4o Mini Audio13.1%
14Kimi Audio 7B10.4%
15Lfm2 Audio 1 5B9.3%

15 of 15

#
1	GPT Realtime 2 xhighOpenAI	48.5%	–	2026-05-07
2	Tml Interaction SmallUnknown	43.4%	–	2026-05-11
3	GPT Realtime 2OpenAI	37.6%	–	2026-05-07
4	Gemini 3.1 Flash Live thinkingGoogle	36.1%	–	2026-03-26
5	GPT Realtime 1.5OpenAI	34.7%	–	2025-12-17
6	Gemini 3.1 Flash LiveGoogle	26.8%	–	2026-03-26
7	Qwen3 Omni 30B A3B InstructAlibabaopen ↗	24.3%	39.4	2026-02-23
8	GPT 4o AudioOpenAI	23.2%	–	2025-12-17
9	Gemini 2.5 Flash Native Audio 12 2025 thinkingGoogle	21.5%	–	2026-02-25
10	GPT RealtimeOpenAI	20.4%	–	2025-12-17
11	GPT Realtime MiniOpenAI	13.9%	–	2026-02-23
12	Gemini 2.5 Flash Native Audio 12 2025 no reasoningGoogle	13.9%	–	2026-03-14
13	GPT 4o Mini AudioOpenAI	13.1%	–	2026-02-23
14	Kimi Audio 7BMoonshot AI	10.4%	–	2025-12-17
15	Lfm2 Audio 1 5BLiquid AI	9.3%	–	2025-12-17

Cite as: BenchLeader, “AudioMC (audio output) leaderboard”, https://www.benchleader.com/benchmarks/scale_audiomc_audio, data as of 19 Sept 2026.

AudioMC (audio output): questions

What does AudioMC (audio output) measure?: The AudioMC questions with the model required to answer in spoken audio. Scores are reported in percent of tasks solved; higher is better.
Which AI model leads AudioMC (audio output)?: GPT Realtime 2 leads AudioMC (audio output) with 48.5% as of 19 Sept 2026, ahead of Tml Interaction Small at 43.4%.
How many models have AudioMC (audio output) results?: 15 model configurations have a AudioMC (audio output) result on BenchLeader, all taken from Scale AI SEAL.
Who runs AudioMC (audio output) and how often is it updated?: AudioMC (audio output) is published by Scale AI SEAL. BenchLeader re-reads the published results every morning and records the date each result was published.
Does AudioMC (audio output) count toward the BenchLeader Index?: No. AudioMC (audio output) is shown for reference but left out of the composite index.