Author: RadarAI Editorial
Editor: RadarAI Editorial
Last updated: 2026-10-02
Review status: Editorial review pending
Brief
速报
官方
AI动态
开源
OpenAI launched the ChatGPT virtual try-on feature globally, while terminating collaboration with three researchers due to improper handling of sensitive information; Microsoft released its first real-time streaming speech transcription model, MAI-Transcribe-2-Streaming, topping the Artificial Analysis leaderboard with a 2.50% word error rate and 0.13-second latency [3][4]. The open-source camp continues to gain momentum, with Qwen-Image-2.1 becoming the top open-weight model on both AA-Image leaderboards...
Editorial standards and source policy: Editorial standards, Team. Content links to primary sources; see Methodology.
## 🔍 Core Insights
**OpenAI** launched the **ChatGPT virtual try-on** feature globally, while terminating collaboration with three researchers due to improper handling of sensitive information; **Microsoft** released its first real-time streaming speech transcription model, **MAI-Transcribe-2-Streaming**, topping the Artificial Analysis leaderboard with a **2.50% word error rate** and **0.13-second latency** [3][4]. The open-source camp continues to gain momentum, with **Qwen-Image-2.1** becoming the top open-weight model on both AA-Image leaderboards, and the **GLM 5.3** series launching on Cursor [9][7].
## 🚀 Key Updates
- **OpenAI launches ChatGPT virtual try-on and product favorites globally** [1]: Based on the ChatGPT Images 2.5 model, it supports uploading photos to try on clothing and accessories.
- **Microsoft releases MAI-Transcribe-2-Streaming real-time speech transcription model** [3]: Supports 60 languages, ranking first with a 2.50% word error rate and 0.13-second latency.
- **Black Forest Labs releases FLUX 3 Image** [2]: Focused on precise editing and 4K support, with an open-weight plan to be released in the coming weeks.
- **OpenAI terminates collaboration with three researchers** [4]: Due to bypassing procedures and providing company confidential information to a third-party AI safety organization.
- **Qwen-Image-2.1 tops AA-Image open-weight model leaderboard** [7]: Ranked 18th on both the T2I and Editing leaderboards, first among open-source models.
- **GLM 5.3 series launches on Cursor** [9]: GLM 5.3 Max becomes the highest-scoring open-weight model on CursorBench 4.0.
- **Musk cuts Tesla AI5/AI6 chip memory usage** [1]: AI5 RAM reduced from 144GB to 72GB, reallocating resources for Optimus robot mass production.
- **Perplexity open-sources multimodal decision model pplx-decider-27b** [23]: Launches Decisions API, with input token pricing at 4 cents per million and output free.
## 🔗 Sources
[1] OpenAI launches virtual try-on and product favorites globally on ChatGPT — https://aihot.news/items/bmu5q5uch58h5bmlhglxe3ygv
[2] Black Forest Labs releases FLUX 3 Image, supporting precise editing and 4K — https://aihot.news/items/go77pg70gyp485d3nwd66kbez
[3] Microsoft releases first real-time streaming speech transcription model MAI-Transcribe-2-Streaming, topping Artificial Analysis with 2.50% word error rate and 0.13-second latency — https://aihot.news/items/r28hmqp7e3viwbp0r68yokncb
[4] OpenAI terminates collaboration with three researchers for violating sensitive information access and sharing rules — https://aihot.news/items/qugld236jgo61cwymxs7o5v25
[7] Artificial Analysis evaluation: Qwen-Image-2.1 tops both AA-Image leaderboards among open-weight models — https://aihot.news/items/kaz49j9sbrf1rcugch9d2fdjm
[9] GLM 5.3 series launches on Cursor — https://aihot.news/items/ga75rs11tfy9u96f3ay6xuzjt
[23] Perplexity open-sources multimodal decision model pplx-decider-
OpenAI launched the ChatGPT virtual try-on feature globally, while terminating collaboration with three researchers due to improper handling of sensitive information; Microsoft released its first real-time streaming speech transcription model, MAI-Transcribe-2-Streaming, topping the Artificial Analysis leaderboard with a 2.50% word error rate and 0.13-second latency [3][4]. The open-source camp continues to gain momentum, with Qwen-Image-2.1 becoming the top open-weight model on both AA-Image leaderboards, and the GLM 5.3 series launching on Cursor [9][7].
🚀 Key Updates
- OpenAI launches ChatGPT virtual try-on and product favorites globally [1]: Based on the ChatGPT Images 2.5 model, it supports uploading photos to try on clothing and accessories.
- Microsoft releases MAI-Transcribe-2-Streaming real-time speech transcription model [3]: Supports 60 languages, ranking first with a 2.50% word error rate and 0.13-second latency.
- Black Forest Labs releases FLUX 3 Image [2]: Focused on precise editing and 4K support, with an open-weight plan to be released in the coming weeks.
- OpenAI terminates collaboration with three researchers [4]: Due to bypassing procedures and providing company confidential information to a third-party AI safety organization.
- Qwen-Image-2.1 tops AA-Image open-weight model leaderboard [7]: Ranked 18th on both the T2I and Editing leaderboards, first among open-source models.
- GLM 5.3 series launches on Cursor [9]: GLM 5.3 Max becomes the highest-scoring open-weight model on CursorBench 4.0.
- Musk cuts Tesla AI5/AI6 chip memory usage [1]: AI5 RAM reduced from 144GB to 72GB, reallocating resources for Optimus robot mass production.
- Perplexity open-sources multimodal decision model pplx-decider-27b [23]: Launches Decisions API, with input token pricing at 4 cents per million and output free.
🔗 Sources
[1] OpenAI launches virtual try-on and product favorites globally on ChatGPT — https://aihot.news/items/bmu5q5uch58h5bmlhglxe3ygv
[2] Black Forest Labs releases FLUX 3 Image, supporting precise editing and 4K — https://aihot.news/items/go77pg70gyp485d3nwd66kbez
[3] Microsoft releases first real-time streaming speech transcription model MAI-Transcribe-2-Streaming, topping Artificial Analysis with 2.50% word error rate and 0.13-second latency — https://aihot.news/items/r28hmqp7e3viwbp0r68yokncb
[4] OpenAI terminates collaboration with three researchers for violating sensitive information access and sharing rules — https://aihot.news/items/qugld236jgo61cwymxs7o5v25
[7] Artificial Analysis evaluation: Qwen-Image-2.1 tops both AA-Image leaderboards among open-weight models — https://aihot.news/items/kaz49j9sbrf1rcugch9d2fdjm
[9] GLM 5.3 series launches on Cursor — https://aihot.news/items/ga75rs11tfy9u96f3ay6xuzjt
[23] Perplexity open-sources multimodal decision model pplx-decider-
← Back to Updates