Library
Library/AI Model Releases
EVENTS708
ARTICLES292
CATEGORYAI
Active

New AI model releases, benchmarks, and capabilities. Foundation models, open-source releases, and API launches from leading AI labs.

Events Articles
LATEST EVENTS
Entity Type Summary Impact Confidence
Google product Google released the Switch Transformer, a massive model with 1.6 trillion parameters. medium 90%
Google product Google released GShard, the first Mixture of Experts model at massive scale with 600B parameters for machine translation… medium 95%
OpenAI product OpenAI released GPT-3, a 175B parameter model demonstrating few-shot learning and sparking the modern LLM revolution. high 95%
Google product Google released T5, a unified text-to-text transfer transformer framework for all NLP tasks. high 95%
Meta product Meta released RoBERTa, an optimized version of BERT that achieved new state-of-the-art results with improved training. medium 95%
Google / CMU product Google and CMU released XLNet, an autoregressive pretraining model that outperformed BERT on 20 NLP tasks. medium 95%
OpenAI product OpenAI released GPT-2, a model initially withheld due to misuse concerns, demonstrating emergent text generation quality… high 95%
Google product Google released BERT, a model that revolutionized NLP benchmarks and became foundational for search engines. high 95%
OpenAI product OpenAI released GPT-1, its first generative pre-trained transformer model for language understanding. high 95%
AllenAI product AllenAI released ELMo, a model providing contextualized word representations using bidirectional LSTMs. medium 95%
Google product Google introduced the Transformer architecture, a foundational component for all modern large language models. high 95%
CallSphere AI announcement CallSphere AI published an in-depth analysis comparing top foundation models like GPT-4, Claude, Gemini, Llama, and Mist… medium 95%
xAI product Grok 4.6 model is mentioned, suggesting a new version release or update for xAI's AI model. medium 80%
OpenAI product ChatGPT is reported to be ultrafast, indicating a significant performance improvement for the model. medium 80%
Meta product Meta's Llama 4 Scout and Maverick were highlighted by Faros AI as large multimodal open-weight models. low 80%
MiniMax product MiniMax's M2.5 and M3 models were highlighted by Faros AI for their low-cost and high-throughput capabilities. low 80%
Moonshot AI product Moonshot AI's Kimi K2 was highlighted by Faros AI as an open-weight model supporting multimodal and multi-agent workflow… low 80%
Alibaba product Alibaba's Qwen 3 was highlighted by Faros AI as a flexible open-weight foundation model for local and cloud use. low 80%
Z.ai product Z.ai GLM-5.2 was highlighted by Faros AI as an open-weight model specifically designed for long engineering tasks. low 80%
DeepSeek product DeepSeek V4 was highlighted by Faros AI as an efficient, high-performance open-weight model for coding. low 80%
Subscribe to this Dataset
Access the data, set alerts, and customize delivery.
Sign up free
WATCHER CONFIG
query: AI model release benchmark GPT Claude Gemini Llama open source foundation model
event_types: ['product', 'announcement', 'partnership']
impact_floor: all
time_window: 7
Export CSV Export JSON
AI Model Releases
AI · Updated Apr 04, 2026
ACTIVE
MONITORING OBJECTIVE
New AI model releases, benchmarks, and capabilities. Foundation models, open-source releases, and API launches from leading AI labs.
NOTIFICATION FREQUENCY
Real-time notifications based on signal detection
SIGNAL METRICS
708
Total events
Apr 04
Last activity
DATA SCHEMA
Model string
Lab string
Parameters string
Benchmark string
Access string
Signal event details
AI Model Releases
FIELD DATA
RAW DATA