option

Discover quality AI tools

Bring together the world’s leading artificial intelligence tools to help improve work efficiency

Author AlbertDavis

Articles published by AlbertDavis

A total of 6 articles
August 4, 2026

Tencent Hunyuan launched Hy ASR3.0 preview, integrating large language model Hy3 for context-aware speech recognition. Word error rates on open-source sets: Mandarin 3.34%, English 2.62%, Cantonese 3.12%. Self-built evaluations cover dialects, professional terms, and noisy conditions, achieving industry-leading performance.

Tencent Hunyuan launched Hy ASR3.0 preview, integrating large language model Hy3 for context-aware speech recognition. Word error rates on open-source sets: Mandarin 3.34%, English 2.62%, Cantonese 3.12%. Self-built evaluations cover dialects, professional terms, and noisy conditions, achieving industry-leading performance.

Tencent Hunyuan launched Hy ASR3.0 preview, integrating large language model Hy3 for context-aware speech recognition. Word error rates on open-source sets: Mandarin 3.34%, English 2.62%, Cantonese 3.12%. Self-built evaluations cover dialects, professional terms, and noisy conditions, achieving industry-leading performance.
July 31, 2026

ByteDance Seed launched Seedance 2.5, a video generation model supporting 30-second single outputs with multi-round extension. It enables up to 30 images, 10 videos, and 10 audio references, plus post-editing via timestamps. Now available on Jimeng AI and Doubao Pro, with API coming soon.

ByteDance Seed launched Seedance 2.5, a video generation model supporting 30-second single outputs with multi-round extension. It enables up to 30 images, 10 videos, and 10 audio references, plus post-editing via timestamps. Now available on Jimeng AI and Doubao Pro, with API coming soon.

ByteDance Seed launched Seedance 2.5, a video generation model supporting 30-second single outputs with multi-round extension. It enables up to 30 images, 10 videos, and 10 audio references, plus post-editing via timestamps. Now available on Jimeng AI and Doubao Pro, with API coming soon.
June 22, 2026

Zhipu (02513.HK) surged over 40% on June 22, hitting 2980 HKD and breaking 10 billion HKD market cap, up 2000% since January. The rally followed US restrictions on Anthropic models; Zhipu quickly open-sourced its GLM-5.2 model, which topped Code Arena and integrated with domestic computing platforms. API usage grew 400% despite price hikes. The company plans A-share listing on STAR Market to raise 1.5 billion yuan. Despite losses, market confidence remains strong.

Zhipu (02513.HK) surged over 40% on June 22, hitting 2980 HKD and breaking 10 billion HKD market cap, up 2000% since January. The rally followed US restrictions on Anthropic models; Zhipu quickly open-sourced its GLM-5.2 model, which topped Code Arena and integrated with domestic computing platforms. API usage grew 400% despite price hikes. The company plans A-share listing on STAR Market to raise 1.5 billion yuan. Despite losses, market confidence remains strong.

Zhipu (02513.HK) surged over 40% on June 22, hitting 2980 HKD and breaking 10 billion HKD market cap, up 2000% since January. The rally followed US restrictions on Anthropic models; Zhipu quickly open-sourced its GLM-5.2 model, which topped Code Arena and integrated with domestic computing platforms. API usage grew 400% despite price hikes. The company plans A-share listing on STAR Market to raise 1.5 billion yuan. Despite losses, market confidence remains strong.
June 1, 2026

Chinese AI firm Xiyu Technology released MiniMax M3, an open-source model with 1M-token context, native multimodal support (image, video, desktop ops), and top SWE-Bench Pro score of 59.0%, surpassing GPT and Gemini. Its sparse attention architecture reduces computation by half, boosting speed 9x in understanding and 15x in generation. API live; weights and report to be open-sourced within 10 days.

Chinese AI firm Xiyu Technology released MiniMax M3, an open-source model with 1M-token context, native multimodal support (image, video, desktop ops), and top SWE-Bench Pro score of 59.0%, surpassing GPT and Gemini. Its sparse attention architecture reduces computation by half, boosting speed 9x in understanding and 15x in generation. API live; weights and report to be open-sourced within 10 days.

Chinese AI firm Xiyu Technology released MiniMax M3, an open-source model with 1M-token context, native multimodal support (image, video, desktop ops), and top SWE-Bench Pro score of 59.0%, surpassing GPT and Gemini. Its sparse attention architecture reduces computation by half, boosting speed 9x in understanding and 15x in generation. API live; weights and report to be open-sourced within 10 days.
March 25, 2026

OpenAI is shutting down its video social media app Sora just six months after launch. The app, which generated realistic digital avatars, failed due to plummeting downloads, a collapse of its content ecosystem filled with deepfakes and copyright violations, and mounting legal pressures. OpenAI will now focus on integrating its advanced Sora2 model as a tool within ChatGPT.

OpenAI is shutting down its video social media app Sora just six months after launch. The app, which generated realistic digital avatars, failed due to plummeting downloads, a collapse of its content ecosystem filled with deepfakes and copyright violations, and mounting legal pressures. OpenAI will now focus on integrating its advanced Sora2 model as a tool within ChatGPT.

OpenAI is shutting down its video social media app Sora just six months after launch. The app, which generated realistic digital avatars, failed due to plummeting downloads, a collapse of its content ecosystem filled with deepfakes and copyright violations, and mounting legal pressures. OpenAI will now focus on integrating its advanced Sora2 model as a tool within ChatGPT.
September 26, 2025

Tech startups now stay private longer (14 yrs vs 5 in 1999), with top private US tech firms worth $1.8T. Private markets hit $15T AUM as new VC models (1.85% fees, SEC oversight) democratize access to late-stage growth like Stripe and OpenAI amid $20T AI opportunity. The House passed a 2025 bill expanding private market access beyond wealthy investors.

Tech startups now stay private longer (14 yrs vs 5 in 1999), with top private US tech firms worth $1.8T. Private markets hit $15T AUM as new VC models (1.85% fees, SEC oversight) democratize access to late-stage growth like Stripe and OpenAI amid $20T AI opportunity. The House passed a 2025 bill expanding private market access beyond wealthy investors.

Tech startups now stay private longer (14 yrs vs 5 in 1999), with top private US tech firms worth $1.8T. Private markets hit $15T AUM as new VC models (1.85% fees, SEC oversight) democratize access to late-stage growth like Stripe and OpenAI amid $20T AI opportunity. The House passed a 2025 bill expanding private market access beyond wealthy investors.
OR