Mega-ASR: Open-Source Model Cuts Noise-Induced Hallucinations by 30%
DeepSeek is making its 75% discount on V4-Pro API permanent after May 31, 2026, slashing cache-hit, cache-miss, and output prices to one-quarter of original rates.
Qwen
Most Recent
Nous Research Proves Tokenizer-Free LLMs Are Possible
Tencent open-sources its Hy-MT2 translation models, with a compact 440MB version that can run offline on phones. The models support 33 languages and 5 Chinese dialects, and the Hy Translate app is now available.
Qwen3.7-Max API Launches on Alibaba Cloud, Costs 12 Yuan per Million Tokens
Tencent open-sources its Hy-MT2 translation models, with a compact 440MB version that can run offline on phones. The models support 33 languages and 5 Chinese dialects, and the Hy Translate app is now available.
Zhipu's GLM-5.1 Turbo API Hits 400 Tokens/s, Claims Global Speed Record
Tencent open-sources its Hy-MT2 translation models, with a compact 440MB version that can run offline on phones. The models support 33 languages and 5 Chinese dialects, and the Hy Translate app is now available.
Alibaba Cloud Launches WonderClip, a Full-Stack AI Video Creation Platform
Tencent open-sources its Hy-MT2 translation models, with a compact 440MB version that can run offline on phones. The models support 33 languages and 5 Chinese dialects, and the Hy Translate app is now available.
Draft Models Give Way to MTP Speculative Decoding on Local Hardware
Tencent open-sources its Hy-MT2 translation models, with a compact 440MB version that can run offline on phones. The models support 33 languages and 5 Chinese dialects, and the Hy Translate app is now available.
Alibaba's first consumer AI hit Miao Duck Camera disbands team, stops updates
Tencent's new Chronicles-OCR benchmark reveals that top AI models like GPT-5 and Gemini fail miserably at recognizing ancient Chinese scripts, scoring near zero on oracle bone and bronze inscriptions.
Alibaba’s Qwen3.5-LiveTranslate reads video and clones voices in real time
Tencent's new Chronicles-OCR benchmark reveals that top AI models like GPT-5 and Gemini fail miserably at recognizing ancient Chinese scripts, scoring near zero on oracle bone and bronze inscriptions.
Qwen3.7-Max Autonomously Wrote 1,158 Lines of Code to 10x Chinese Chip Performance
Tencent's new Chronicles-OCR benchmark reveals that top AI models like GPT-5 and Gemini fail miserably at recognizing ancient Chinese scripts, scoring near zero on oracle bone and bronze inscriptions.
Microsoft's ECHO makes CLI agents learn by predicting terminal output
Tencent's new Chronicles-OCR benchmark reveals that top AI models like GPT-5 and Gemini fail miserably at recognizing ancient Chinese scripts, scoring near zero on oracle bone and bronze inscriptions.