SpaceXAI's newest speech-to-text model ranks first among 32 streaming models on Artificial Analysis, at the same price as version 1.0.
xAI is introducing Grok Voice Transcribe 2.0, its latest speech-to-text model, and says it is twice as accurate as Grok Voice Transcribe 1.0 at the same price. The September 18, 2026 release is built on the audio foundation model behind Grok Voice, which already powers tens of thousands of customer-support calls a day, transcribes millions of hours of video narration, and runs voice agents in physical products, including the Grok assistant in Tesla vehicles.
Most transcription models do well on clean, single-speaker audio. Real-world audio is harder: flaky phone lines, competing voices, local accents, and phone numbers or emails read aloud. On the public Artificial Analysis leaderboard, Grok Voice Transcribe 2.0 ranks first for accuracy among 32 streaming models. Internal tests on production traffic — telephony support calls, conversations with Grok, spoken credentials, and short multilingual voice commands — show gains versus version 1.0 on all four sets. On telephony, xAI says the new model leads every system it tested. Multilingual accuracy is the largest jump: the model transcribes dozens of languages, detects language automatically, and follows mid-recording switches in a single pass. On a short-phrase set with little context, word error rate drops from 20.6% to 6.8%.
Existing Speech-to-Text API integrations get the accuracy improvement with no code changes. Features include batch and streaming transcription, word-level timestamps, speaker diarization at no extra cost, up to eight channels, key-term biasing for domain vocabulary, written-form numbers and emails, filler-word removal, and smart turn detection for voice agents. Atlassian found the model more accurate than its previous solution for Loom videos, opening workflows that send a recording into Cursor for code updates. Pricing is unchanged: batch transcription is $0.10 per hour of audio and streaming is $0.20 per hour, with diarization, timestamps, and key terms included. Grok Voice Transcribe 2.0 will soon become the default in the Speech-to-Text API, and version 1.0 will be deprecated in the coming weeks. Teams that need the older model during the transition can pin grok-voice-transcribe-1.0. Read the full announcement on the xAI news post.
Disclaimer
This is an AI-summarized article from authoritative sources. If you want your content removed, please .
