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.
SEARCH
Search English reports
English reports are listed here. AI-generated drafts are clearly labeled until an editor reviews them.
AI-generated · reviewedImpact 94/100
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.
Manus launches a cloud-based general AI agent
Manus asks users for an outcome, then plans, researches, codes, operates websites, and returns finished artifacts from a cloud virtual machine.
DeepSeek tops the U.S. App Store and jolts AI markets
DeepSeek passes ChatGPT on the U.S. App Store's free chart as Nvidia loses roughly $593 billion in market value during an AI-sector selloff.
OpenAI previews Operator, an agent that controls a browser
Operator uses a visual computer-use model to click, type, and scroll through web tasks, handing control back for logins, payments, and sensitive actions.
DeepSeek releases the open-weight reasoning model R1
DeepSeek publishes R1, R1-Zero, six distilled models, and its reinforcement-learning recipe, pushing open reasoning models into the frontier conversation.
DeepSeek launches a free consumer app before R1
DeepSeek puts V3, web search, a reasoning mode, and file analysis into a free mobile app five days before the R1 release.
DeepSeek-V3 brings an efficiency-first open model to the frontier race
DeepSeek releases a 671B MoE model with 37B active parameters, open weights, low API prices, and unusually detailed efficiency disclosures.
DeepSeek previews its reasoning line before V3 ships
R1-Lite-Preview exposes long-form reasoning on the web and promises an open model and API, showing that R1 was already underway before V3's release.
Alibaba expands Qwen2.5 into a broad open-model suite
Qwen2.5 spans small local models through a 72B flagship, with stronger instruction following and specialist code and math editions.
OpenAI previews o1, a model trained to spend more time reasoning
OpenAI releases o1-preview, shifting frontier-model competition toward reinforcement learning and additional computation at inference time.
DeepSeek releases the efficiency-focused DeepSeek-V2
DeepSeek releases a 236B mixture-of-experts model that combines sparse activation with latent attention to cut training and inference costs.
Meta researchers publish multi-token prediction
Meta FAIR researchers propose training language models to predict several future tokens at once, work that later informs DeepSeek-V3's MTP design.
DeepSeekMath introduces the GRPO method later used by R1
DeepSeek releases math models and Group Relative Policy Optimization, creating a direct algorithmic predecessor to R1's reinforcement-learning recipe.
DeepSeek publishes the sparse architecture behind its later models
DeepSeekMoE uses fine-grained experts and shared expert isolation to improve sparse-model efficiency, with 16B weights and training code released.
DeepSeek starts with an open-weight code model
DeepSeek releases code models from 1.3B to 33B parameters, making open weights and code part of its product strategy from the outset.
Zhipu AI releases ChatGLM3 with native tool use
ChatGLM3 adds native function calling and code execution to a compact 6B model, shifting the ChatGLM line from conversation toward agent-like workflows.
Alibaba Cloud opens the Qwen foundation-model family
Alibaba Cloud releases Qwen as a multilingual open-weight foundation-model family, giving developers a major China-built alternative across model sizes.