AI Battle > Rankings > Speech Reasoning (Big Bench Audio)
null by null leads the Speech Reasoning (Big Bench Audio) ranking with a score of 0.99. The ranking covers 20 evaluated speech-to-speech from a public dataset that AI Battle re-checks daily.
Last updated: . Source: public benchmark datasets and provider pricing, aggregated daily; 20 entries shown.
| # | Entry | Creator | Value | Δ vs #1 |
|---|---|---|---|---|
| 1 | Qwen Audio 3.0 Realtime Plus, Alibaba Cloud | 0.99 | 0.0% | |
| 2 | Qwen3.5 Omni Plus Realtime | 0.99 | -0.5% | |
| 3 | Step-Audio R1.1 (Realtime) | 0.98 | -1.6% | |
| 4 | Grok Voice Think Fast 2.0 High | 0.97 | -2.0% | |
| 5 | Grok Voice Think Fast 1.0 | 0.97 | -2.1% | |
| 6 | Gemini 3.1 Flash Live Preview - High | 0.97 | -2.6% | |
| 7 | GPT-Realtime-2 (High) | 0.97 | -2.6% | |
| 8 | Qwen Audio 3.0 Realtime Flash, Alibaba Cloud | 0.96 | -3.1% | |
| 9 | GPT-Realtime-2.1 High, OpenAI | 0.96 | -3.4% | |
| 10 | Grok Voice Fast 1.0 | 0.93 | -5.9% | |
| 11 | Gemini 2.5 Flash Native Audio Dialog Thinking | 0.91 | -8.6% | |
| 12 | GPT-Live-1 (Astra, medium), OpenAI | 0.90 | -9.2% | |
| 13 | GPT-Live-1 (Sol, low), OpenAI | 0.89 | -10.3% | |
| 14 | Nova 2.0 Sonic (Mar 2026) | 0.88 | -11.2% | |
| 15 | Deepslate Opal | 0.85 | -14.3% | |
| 16 | GPT-Realtime-2.1 Mini High, OpenAI | 0.75 | -24.7% | |
| 17 | GPT-Realtime-2 (Minimal) | 0.72 | -27.6% | |
| 18 | Gemini 3.1 Flash Live Preview - Minimal | 0.71 | -28.1% | |
| 19 | Gemini 2.5 Flash Native Audio Dialog | 0.69 | -30.8% | |
| 20 | GPT Realtime Mini (Oct '25) | 0.64 | -35.9% |
Across the 20 numeric entries in this window the median is 0.92 and the mean is 0.88, so the typical entry sits 0.07 below the leader. The window spans 0.64 to 0.99.
10 of 20 entries are at or above the median, and the top three hold 17% of the combined value — read it against the metric below, since a higher number is not "better" for every dataset.
The source dataset for this family does not publish a provider or creator label per entry, so AI Battle cannot group them: the table above is the complete published list. Where the dataset does carry a label (model, gateway and tool families do), this section names the creators and how many entries each holds.
This ranking carries the metric
bbaScore from the
Speech Reasoning (Big Bench Audio) dataset in the tab::s2s family. Units and the
direction of "better" are not declared by the source dataset, so AI Battle does not invent
them: values are shown exactly as published, and the ordering is the source dataset's own.
Rebuilt on from the
public dataset Speech Reasoning (Big Bench Audio) (family tab::s2s).
The source lists 20 entries and all 20 are shown here. AI Battle requests
the dataset's rows, keeps the ordering the source publishes, and computes only the statistics
on this page: median, mean, spread, the #1–#2 gap, the creator breakdown and the raw-data file.
Nothing on this page is generated text about the entries — the numbers come from the source.
Refresh cadence: every deploy re-pulls the dataset, so this page tracks the source's own update cycle. Corrections: see the methodology page.
Machine-readable copy of this table, with the statistics below and the source provenance: /data/rankings/s2s-speech-reasoning-big-bench-audio.json. Free to reuse with attribution to AI Battle and the source dataset.
null by null currently leads the Speech Reasoning (Big Bench Audio) ranking with a score of 0.99, the highest value in this dataset across Speech-to-Speech. AI Battle aggregates the dataset from its public source and re-checks it daily.
The Speech Reasoning (Big Bench Audio) ranking is rebuilt daily from the upstream public dataset. Scores are the source dataset's own published values, so positions track whatever that source last released — each ranking page names the dataset it comes from.
The top 3 are: 1. null (0.99), 2. null (0.99), 3. null (0.98).