
Audio Understanding Benchmarks Leaderboard 2026
Rankings of the best audio language models on MMAU, MMAU-Pro, and other benchmarks covering speech reasoning, music understanding, and environmental sound identification.
They summarize our coverage. We write it.
Newsletters like this one rebroadcast our headlines - often without the full review, the source reading, or the analysis underneath. Our weekly briefing sends the work they paraphrase, straight from the desk, before they get to it.
Free, weekly, no spam. One email every Tuesday. Unsubscribe anytime.

Rankings of the best audio language models on MMAU, MMAU-Pro, and other benchmarks covering speech reasoning, music understanding, and environmental sound identification.

A data-driven comparison of 12 AI customer support platforms covering pricing models, resolution rates, channel coverage, and helpdesk integrations for 2026.

Google's new Gemini 3.1 Flash TTS hits Elo 1,211 on the Artificial Analysis leaderboard and introduces 200-plus audio tags for mid-sentence voice control, available in preview today via the Gemini API.

Google's Gemini 3.1 Flash Live beats GPT-4 Realtime 1.5 on Scale AI's Audio MultiChallenge and takes Search Live to 200+ countries - but it doesn't lead every benchmark.

Cohere releases its first audio model - a 2B-parameter open-source ASR system beating Whisper Large v3 by 27% on the HuggingFace Open ASR Leaderboard.

Mistral's first open-weights TTS model clones voices from 3 seconds of audio, beats ElevenLabs on price, and arrives with real limitations worth knowing.

Mistral releases Voxtral, a pair of open-weights models covering speech recognition and text-to-speech that undercut OpenAI and ElevenLabs on price.

Tencent open-sources Covo-Audio, a 7B end-to-end audio language model with native full-duplex conversation that beats larger closed models on key benchmarks.

A practical guide to building an AI voice agent using platforms like Vapi, Retell, and LiveKit - covering architecture, setup steps, and cost estimates.

Rankings of the best text-to-speech and speech-to-text AI models by naturalness, accuracy, latency, and pricing.

A developer ported NVIDIA's PersonaPlex 7B speech-to-speech model to native Swift using MLX, running full-duplex conversation on Apple Silicon with no cloud, no Python, and faster-than-real-time inference.

New research reveals no speech AI passes a Turing test, adaptive routing slashes LLM costs 82%, and pseudocode planning transforms agent reliability.