An independent benchmarking organization that measures and publishes model performance, pricing, and speed. It serves as a source of third-party figures rather than self-reported ones.
Current rank (1M)
60
Rank over the last 17 days
2026.09.14 – 2026.09.30 · High 59 · Low 90 · Now 60
ElevenLabs released Eleven v4, a voice model with stronger emotional expression, and its low-latency variant Eleven v4 Turbo. Turbo reaches first speech in about 150 milliseconds, and both models support more than 90 languages.
Google released Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking on September 15. The two voice models run tools in the background mid-conversation and reason while they speak, and Extended Thinking took first place on the Speech to Speech Quality Index with a score of 82.6.
The anonymous model that topped usage charts for a week was Z.ai's new multimodal model. All of its traffic ran on Chinese-made AI chips.
New custom benchmarking platform compares not just quality but cost and time per task
The independent evaluator known for measuring model performance is now preparing a tool developers can use directly
Scores 38 on Artificial Analysis Intelligence Index, sits on the Pareto frontier for performance per parameter
Patch release upgrades the grader model and incorporates the latest τ³-Banking version
That's the last story.
