Grok Voice 3.0 tops Artificial Analysis Speech to Speech Index

Gives voice-agent builders ranked speech-to-speech results, covering agentic Tau Voice, preference, and task success, plus the $0.08 per minute price, to choose between realtime APIs.

• Artificial Analysis scores Grok Voice 3.0 at 86.9 on its Speech to Speech Index, ahead of GPT-Live-1 (Astra, medium) at 83.2 and Gemini 3.8 Live Extended Thinking (High) at 82.6, and 5.6 points above Grok Voice 2.0.
• The index averages four measures: Big Bench Audio reasoning, Tau Voice agentic tasks, arena preference and task success rate. Grok Voice 3.0 leads task success at 94.8% but places third on arena preference (Elo 1070), behind two Gemini Live models.
• On Artificial Analysis's Tau Voice agentic benchmark, Grok Voice 3.0 scores 71.6%, second to GPT-Live-1 (Astra, medium) at 74.5% and well up from 56.5% for Grok Voice 2.0.
• Average time to first audio is 0.81 seconds, faster than the GPT-Live-1 and Gemini 3.8 Live configurations tested (1.24s to 1.35s) but slower than Grok Voice 2.0 at 0.70s. Input audio costs $4.80 per hour, unchanged from 2.0.

• AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
• benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.

• postAA-AgentPerf debuts AMD MI355X results versus NVIDIA Blackwell
• postDeepSeek V4.1 Flash lands on AA-AgentPerf with NVIDIA and AMD results
• postSonnet 5.5 effort levels: max tops Coding Agent Index, xhigh saves 77%
• postCache hit rates explain a 9x provider cost gap for agentic tasks

• postOpenAI’s GPT-Live-1 debuts at #1 on the Artificial Analysis Speech to Speech Index at a score of 81.5 using Astra as itsArtificialAnlys
• postGemini 3.8 Live Tops Speech-to-Speech BenchmarksArtificialAnlys
• postGrok Voice Transcribe 2.0 Tops Streaming ASR Accuracy BenchmarkArtificialAnlys

Checking sign-in…

Loading comments…