xAI releases Grok 4
xAI releases Grok 4, its fourth-generation closed language model, with stronger cross-domain generalization and real-time web search retrieval.
SEARCH
Search English reports
Browse published English reports with original sources and content provenance.
AI-generated
Mistral AI releases Magistral Medium 1.1
An iteration of Magistral Medium with stronger handling and association across complex contexts.
Liquid AI releases LFM2
Liquid AI's second-generation lightweight model built on a dynamic liquid neural-network architecture.
Mistral AI releases Magistral Small 1.1
A small iterative update to the Magistral Small line that fine-tunes instruction-following accuracy.
Baidu releases ERNIE 4.5
An upgrade to Baidu's ERNIE series adopting an MoE architecture for better inference efficiency and generation quality.
Tencent releases Hunyuan A13B
Tencent released an open-source, medium-scale MoE model that balances performance with low deployment costs.
Google DeepMind releases Gemma 3n
Google DeepMind releases a 4-billion-parameter research model designed for low-power devices.
MiniMax releases MiniMax-M1
MiniMax releases a new foundation model that unifies text features and multimodal visual features.
ByteDance releases Seed 1.6
ByteDance updates its closed foundation model for its internal application ecosystem.
OpenAI releases GPT-o3 Pro
An advanced commercial variant that pairs systematic deep reasoning with high-precision text generation.
Mistral AI releases Magistral Medium
Mistral's medium-sized closed API model strengthens logical reasoning and multi-tool execution.
Xiaohongshu releases dots.llm1
Xiaohongshu releases a mixture-of-experts model trained on high-quality data from its image-and-text community platform.
Google DeepMind releases Gemini 2.5 Pro Preview 06-05
Google releases the gemini-2.5-pro-preview-06-05 preview as a developer update before the stable Gemini 2.5 Pro release.
Mistral AI releases Magistral Small
Mistral AI releases a lightweight experimental branch focused on fine-grained instruction control.
Xiaomi releases MiMo VL 7B
Xiaomi releases the SFT and RL versions of the open MiMo VL 7B vision-language model for general visual understanding and multimodal reasoning.
Anthropic makes Claude Code generally available
Alongside the Claude 4 launch, Anthropic moved Claude Code out of research preview to general availability across terminal, IDE, background tasks, and SDK.
Anthropic releases Claude 4 Opus
Anthropic's fourth-generation top-tier flagship, raising the ceiling on multimodal understanding and logical tasks.
Anthropic releases Claude 4 Sonnet
The mid-size member of the Claude 4 family, striking a better balance between response speed and reasoning.
Mistral AI releases Devstral
Mistral AI releases a developer-focused model optimized for long-form programming and code-context analysis.
Google releases Veo 3
Google released Veo 3, bringing native generation of ambient sound, sound effects, and character dialogue to its flagship video model.