Performance results, comparisons, and evaluations of AI models across tasks, datasets, and real-world applications.
Baidu's PaddleOCR has become the most-starred OCR project on GitHub, surpassing Tesseract and marking a decisive shift toward AI-driven document recognition.
Mistral AI released Voxtral TTS, a multilingual speech model generating expressive audio from minimal input. Early evaluations show it outperforming ElevenLabs Flash v2.5 with a 62.8% win rate in human preference tests.
A new AI system, SWE-Vision, is outperforming leading models like GPT-5.2 and Seed-2.0-Pro across multiple visual reasoning benchmarks, highlighting advances in code-driven image analysis.
New evaluations show GLM-5.1 closing in on Claude Opus 4.6 in coding performance, highlighting a shrinking gap between open and proprietary AI models.
OpenAI's GPT-5.4 Design Skill shows limited gains in DesignArena rankings. Claude Opus 4.6 remains the top-performing model with a significant lead.
New data suggests Anthropic's Claude Code is outperforming OpenAI's Codex in developer adoption across key usage and repository metrics.
OpenAI introduced GPT-5.4 mini and nano, offering lower-cost AI models with strong reasoning capabilities. Benchmarks show nano emerging as a leader in cost-efficiency despite trade-offs.
NVIDIA unveils Agentic Variation Operators (AVO), showing measurable performance gains over cuDNN and FlashAttention-4 on Blackwell GPUs.
OpenResearcher scores 54.8% on BrowseComp-Plus, beating GPT-4.1 and Claude models with fully offline operation and no reliance on external APIs.
Baidu's SAMA open-source model rivals Kling-Omni in video editing with semantic anchoring and motion alignment.
Wikipedia reclaims top spot over Reddit in Gen AI referral traffic in February 2026, both platforms hovering near 8 million visits.
Baidu's VEGA-3D uses video diffusion models to boost 3D spatial reasoning in multimodal AI - no explicit geometric supervision needed.
Anthropic surpassed OpenAI in enterprise AI adoption, capturing 73% of new clients in just two months.
Baidu's 4B Qianfan-OCR model scores 93.12 on OmniDocBench, surpassing Gemini 3.1 Pro with a unified OCR and layout analysis approach.
AI trained on 700K citation pairs predicts high-impact research and beats GPT-5.2 in scientific forecasting.
OpenAI's GPT-5.4 Mini tops execution benchmarks with 72.1% on OSWorld, while Gemini 3 Flash leads reasoning at 90.4% GPQA Diamond.
GPT-5.4 mini scores 72.1% on OSWorld-Verified, beating Claude Haiku 4.5's 50.7% in efficiency benchmarks.
OpenSeeker open-source search agent tops Alibaba's Tongyi DeepResearch on BrowseComp-ZH using just 11.7K training samples.
A new framework tests AI embedding models across four memory types, revealing that bigger models don't always win on long-horizon tasks.
Zhipu AI's compact GLM-OCR model ranks first on OmniDocBench, setting a new bar for efficient AI document parsing.
We use cookies to improve your experience on our site and to show you relevant advertising. To find our more, read our privacy policy and cookie policy
Saad Ullah
Usman Salis
Marina Lyubimova
Eseandre Mordi
Victoria Bazir
Peter Smith
