Google releases Gemini 2.5 Flash Native Audio
The Gemini Live API gains native-audio conversation, proactive audio and affective-dialogue capabilities.
SEARCH
Search English reports
Browse published English reports with original sources and content provenance.
AI-generated · awaiting review
Google DeepMind releases Gemini 2.5 Flash
Google releases the lightweight Gemini 2.5 model, focusing on first-token latency.
Technology Innovation Institute releases Falcon H1
The Technology Innovation Institute releases an open foundation-model family with new network layers and experimental routing structures.
OpenAI releases the Codex software-engineering agent
OpenAI released Codex, a software-engineering agent that handles coding tasks in parallel inside a cloud sandbox.
Xiaomi releases MiMo 7B
Xiaomi releases the open MiMo 7B family for mathematics, coding and general reasoning, covering Base, SFT, RL-Zero and RL variants.
Mistral AI releases Mistral Medium 3
Mistral AI releases a medium-scale third-generation model with benchmark performance close to leading closed models.
NVIDIA releases Parakeet-TDT-0.6B-v2
NVIDIA releases a 600M FastConformer-TDT model for high-quality English transcription.
Alibaba releases Qwen3 with hybrid thinking modes
Qwen3 combines fast direct answers and longer reasoning in one open-weight family, spanning dense and MoE models up to a 235B flagship.
Z.ai releases Zhipu GLM-4.1V
Z.ai releases an upgraded multimodal model with stronger visual perception and image-text understanding.
OpenAI brings GPT Image 1 to the API
OpenAI opened ChatGPT's image generation to developers as the gpt-image-1 model in the Images API.
OpenAI releases o3
OpenAI's new-generation reasoning-focused model, strengthening deep math and logical proof.
OpenAI releases o4-mini
A lightweight, low-cost variant that carries complex reasoning and chain-of-thought down to a smaller parameter scale.
OpenAI releases GPT-4.1
A maintenance update to the GPT-4 architecture, with modest gains on everyday tasks and better API cost control.
Meta releases Llama 4
Meta releases its fourth-generation open model family with larger parameter scales and an optimized mixture-of-experts architecture.
Google DeepMind releases Gemini 2.5 Pro
Google releases an intermediate upgrade to Gemini's native multimodal architecture with stronger benchmark performance.
OpenAI releases GPT-4o Transcribe
OpenAI releases a new generation of GPT-4o-based speech-to-text models.
NVIDIA releases Nemotron Ultra
NVIDIA releases an open mixture-of-experts model optimized for high-throughput large-scale data processing.
Cohere releases Command A
Cohere releases a new model for enterprise automation workflows and deep API integration.
Google DeepMind releases Gemma 3
Google DeepMind releases a new Gemma family generation that adds native multimodal support to the lightweight open-model tier.
Mistral AI releases Mistral Small 3.1
Mistral AI releases a minor Mistral Small update that fixes known issues and refines selected features.