SpaceXAI launched Grok Voice Transcribe 2.0, a speech-to-text API it says is twice as accurate as version 1.0 at the same price. The model handles noisy phone lines, competing voices, accents, and spoken credentials in batch and streaming modes. It supports 25 languages and auto-detects language switching mid-recording. Batch transcription costs $0.10 per hour and streaming costs $0.20 per hour. SpaceXAI reports it ranks first among 32 streaming models on Artificial Analysis.
Key Takeaways
- SpaceXAI launched Grok Voice Transcribe 2.0, claiming double the accuracy of version 1.0 at the same price
- The model supports 25 languages and auto-detects language switching mid-recording
- Batch transcription costs $0.10/hour and streaming costs $0.20/hour
- SpaceXAI says it ranks first among 32 streaming models on Artificial Analysis
- Alibaba's Qwen team launched Qwen3.8-Omni-Flash, an omni-modal model with a 1M token context window accepting text, images, audio, and video
- Qwen3.8-Omni-Flash shows over 25% average improvement across 29 evaluations versus Qwen3.5-Omni-Plus
- Audio input costs over 98% less per hour compared to the prior model
- AWS published a guide for deploying Hugging Face models on Amazon SageMaker AI using coding agents like Kiro
- The guide deploys Qwen/Qwen3-0.6B to an ml.g5.xlarge instance in US East (N. Virginia)
- Lawmakers compare AI fears to the Y2K bug, which cost about $100 billion to fix; Anthropic's Dario Amodei warns AI could cause economic problems while President Trump has called risk warnings a hoax
SpaceXAI Launches Grok Voice Transcribe 2.0 With Double Accuracy
SpaceXAI released Grok Voice Transcribe 2.0, a speech-to-text API it says is twice as accurate as version 1.0 at the same price. The model handles noisy phone lines, competing voices, accents, and spoken credentials in batch and streaming modes. It supports 25 languages and auto-detects language switching mid-recording. Batch transcription costs $0.10 per hour and streaming costs $0.20 per hour. SpaceXAI reports it ranks first among 32 streaming models on Artificial Analysis.
Alibaba Qwen Releases Omni-Modal Qwen3.8-Omni-Flash Model
Alibaba's Qwen team launched Qwen3.8-Omni-Flash, an omni-modal model with a 1M token context window that accepts text, images, audio, and video. It focuses on agentic audio-video understanding and tool use, processing only the most relevant parts of long videos. Qwen reports accuracy gains on OmniVideoBench and over 25% average improvement across 29 evaluations versus Qwen3.5-Omni-Plus. Audio input costs over 98% less per hour compared to the prior model. The model is available as a hosted API on QwenCloud, Alibaba Cloud Model Studio, and Qwen Studio.
New Jersey Lawmakers Discuss Artificial Intelligence Concerns on CBS
New Jersey Rep. Josh Gottheimer and Brooklyn Sen. Andrew Gounardes appeared on CBS News New York's The Point with Marcia Kramer to discuss concerns about artificial intelligence. The full episode aired on September 19, 2026. The two politicians addressed potential risks and oversight needs for AI technology.
Rep. Gottheimer Proposes AI Guardrails With Bipartisan Bill
New Jersey Rep. Josh Gottheimer introduced a bipartisan bill calling for artificial intelligence guardrails. He partnered with Republican New York Rep. Mike Lawler on the legislation. Gottheimer said AI offers great opportunities across multiple fields but warned that threats could outweigh benefits if it is not regulated properly.
AWS Guide Shows Deploying Hugging Face Models on SageMaker AI
Amazon Web Services published a guide for deploying Hugging Face models on Amazon SageMaker AI using coding agents like Kiro. The process involves six phases covering AWS context discovery, environment setup, role creation, container selection, endpoint creation with autoscaling and CloudWatch alarms, and smoke testing. Users need an AWS account, AWS CLI v2, Python 3.10 to 3.12, and a coding agent that supports skills. The guide deploys Qwen/Qwen3-0.6B to an ml.g5.xlarge instance in US East (N. Virginia).
Sen. Rounds outlines bipartisan AI legislation on national program
Sen. Mike Rounds, R-S.D., discussed congressional legislation focused on AI innovation and regulation. He appeared on the program Life, Liberty and Levin to explain the bills. The legislation aims to address how the government should handle artificial intelligence.
AI fears draw Y2K comparisons as lawmakers push for rules
Lawmakers from both parties are worried about AI and compare it to the Y2K bug. The US spent about $100 billion on Y2K fixes back then. Tech leaders like Anthropic's Dario Amodei warn that AI could cause economic problems. Former worker Jacob Coxon says AI could be very dangerous by the end of the decade. President Trump has called the risk warnings a hoax and wants fast growth to beat China.
Morningstar launches AI portfolio tool Lumonic 12.0
Morningstar's PitchBook unit launched Lumonic 12.0 in September 2026. The AI-powered platform helps institutional investors track portfolio governance and data lineage. A new survey found that 71% of asset owners believe AI will significantly impact their portfolios in two years. About 55% expect AI to improve decisions, but 45% fear it will raise investment risks.
Gov. Spanberger unveils Virginia data center and AI plan
Gov. Abigail Spanberger unveiled Virginia's Data Center Accountability Framework in Richmond. She also signed an executive order creating a new AI Task Force. The plan aims to give communities a voice, protect against high energy bills, and set strong environmental standards. It bans NDAs for data center projects and requires local approval for projects using more than 25 megawatts.
China central bank adviser warns AI may worsen economy gap
Huang Yiping, a People's Bank of China adviser, said AI could make China's weak demand and strong supply problem worse. He spoke at an economic forum in Beijing on September 19. He called for market reforms and more borrowing to help local governments and companies. He also urged China to focus more on domestic consumption instead of only exports.
Suno's New AI Music Model Faces Lawsuit From Sony and Universal
Suno released a new AI music model called v6 that was trained with permission from some record labels. However, Sony and Universal were not part of those deals. These two companies are suing Suno because they believe the v6 model still uses data from earlier models that were trained on copyrighted songs without permission. Suno says the lawsuit is flawed and that its new model was trained on licensed content and community feedback.
Sources
- SpaceXAI Releases Grok Voice Transcribe 2.0: A Speech-to-Text API Claiming 2x Accuracy Over 1.0 at $0.10 per Hour
- Alibaba Qwen Releases Qwen3.8-Omni-Flash: A 1M-Context Omni-Modal Model Built Around Agentic Audio-Video Understanding and Tool Use
- Artificial intelligence concerns | The Point Full Episode 9.19.26
- N.J. Rep. Josh Gottheimer calls for artificial intelligence guardrails with bipartisan bill
- Deploy Hugging Face models on Amazon SageMaker AI with coding agents
- A closer look at bipartisan legislation focused on AI
- AI concerns beginning to mirror Y2K-era as push for regulation continues
- Morningstar (MORN) Puts AI Portfolio Tools And Asset Owner Shifts In Focus
- Governor Spanberger holds news conference on data centers and AI
- China central bank adviser says AI could deepen supply-demand imbalance
- Suno's Latest AI Model Got Some Label Support. But Not All of Them
Comments
Please log in to post a comment.