BREAKING: SpaceXAI has launched Grok Voice Transcribe 2.0, delivering twice the accuracy of its previous model at the same price.

• Ranks #1 for accuracy among 32 streaming models on Artificial Analysis
• Transcribes dozens of languages
• Automatically detects the language being spoken
• Handles language switching within the same recording
• Transcribes uploaded files, URLs and live audio streams
• Adds precise timestamps and confidence scores for every word
• Separates and labels different speakers at no extra cost
• Transcribes up to 8 audio channels independently
• Lets users add up to 100 key terms, including product names and medical words
• Detects when someone has finished speaking, making voice agents feel more natural
• Handles noisy calls, different accents, overlapping voices, phone numbers and email addresses
• Existing API integrations receive the accuracy upgrade without any code changes
Its error rate for short voice commands dropped from 20.6% to just 6.8%.

Pricing remains incredibly low:
• Batch transcription: $0.10 per hour
• Real-time streaming: $0.20 per hour

Atlassian found it more accurate than its previous solution and now uses it to transcribe every Loom video.

Grok Voice Transcribe 2.0 will soon become the default model in xAI’s Speech-to-Text API.