SenseTime launches SenseNova V6.5 at WAIC 2025
SenseTime launched SenseNova V6.5 at a WAIC 2025 forum on 31 July 2025, along with an agent upgrade for its Raccoon assistant.
Total 671 · Clear filters
AI21 Labs releases Jamba 1.7
In July 2025 AI21 Labs published Jamba Large 1.7 and Jamba Mini 1.7, updates to its Jamba model family that focus on grounding answers in the supplied context and on following instructions more reliably.
Lovable rolls out Agent Mode in beta
On 30 June 2025 Lovable began rolling out Agent Mode in beta, a mode that lets its builder search its own codebase, read files, debug from logs and network activity, and use the web instead of only answering prompts.
Baidu open-sources the ERNIE 4.5 model family under Apache 2.0
On 30 June 2025 Baidu open-sourced ten ERNIE 4.5 models under the Apache 2.0 license, ranging from 0.3B to 424B parameters, reversing its earlier closed approach.
Tencent open-sources Hunyuan-A13B
On 27 June 2025 Tencent open-sourced Hunyuan-A13B on Hugging Face with a technical report, in fast and slow thinking modes and several quantized versions.
Google releases the open-source Gemini CLI
On 25 June 2025 Google released Gemini CLI, an open-source agent for the terminal, with a free tier for personal Google accounts that Google described as the industry's largest allowance.
Midjourney launches its first video model
On 18 June 2025, Midjourney released V1, its first video model, which turns images into short clips and names real-time open-world simulation as a long-term goal.
ByteDance Seed releases the Seedance 1.0 video model
ByteDance Seed published the Seedance 1.0 technical report on 10 June 2025 and announced the model on 11 June. It generates 1080p video from text or images and ranked first on two Artificial Analysis leaderboards.
Volcano Engine launches Doubao 1.6 (Seed1.6) at its FORCE conference
On 11 June 2025 Volcano Engine released Doubao Large Model 1.6, known as Seed1.6, a mixture-of-experts model with a 256K context window and a Thinking variant that scored 683 on the humanities paper of the 2025 Shandong gaokao.
OpenAI releases o3-pro
On 10 June 2025 OpenAI released o3-pro, the highest-compute reasoning option in its lineup, aimed at questions where reliability matters more than speed and offered to ChatGPT Pro subscribers and through the API.
Mistral AI releases Magistral, its first reasoning models
Mistral AI released Magistral Small and Magistral Medium on 10 June 2025, its first reasoning models. Medium scored 73.6% on AIME 2024, and Small, a 24B model, shipped under Apache 2.0.
ModelBest releases MiniCPM4 for end devices
On 9 June 2025 the MiniCPM team released MiniCPM4, a language model built for end-side devices, with open weights and an inference stack combining sparse attention, ternary quantization and speculative sampling.
Apple opens its on-device model to developers with second-generation foundation models
On 9 June 2025, at WWDC, Apple introduced its second-generation foundation models and opened the on-device model to third-party apps through the Foundation Models framework, which shipped with iOS 26 in September.
MiniMax releases MiniMax-M1, an open reasoning model with 1M context
In June 2025 MiniMax released MiniMax-M1, an open reasoning model with a native 1-million-token context. It has 456B total parameters and was released under Apache 2.0, and its design was compared with DeepSeek R1 in the release figures.
EleutherAI releases Common Pile v0.1, an openly licensed 8 TB dataset
In June 2025 EleutherAI and its partners released Common Pile v0.1, an 8 TB collection of public-domain and openly licensed text for training language models. The team also used it to train two 7B models.
Black Forest Labs introduces FLUX.1 Kontext
On 29 May 2025 Black Forest Labs introduced FLUX.1 Kontext, a model family for editing existing images from instructions, released in closed and open-weight forms.
Sarvam AI open-sources an Indic reasoning model built on Mistral Small
On 23 May 2025 Sarvam AI released Sarvam-M, an open-weights Indic model built on Mistral Small, publishing the weights, a playground and APIs at the same time.
Anthropic launches Claude Opus 4 and Claude Sonnet 4
On 22 May 2025 Anthropic launched Claude Opus 4 and Claude Sonnet 4, hybrid models with extended thinking and tool use. Anthropic called Opus 4 the world's best coding model, and Claude Code became generally available the same day.
Mistral AI and All Hands AI release Devstral
On 21 May 2025 Mistral AI and All Hands AI released Devstral, an openly licensed coding model built to work as an agent on real software issues, scoring 46.8% on SWE-Bench Verified.
Google launches Veo 3 with native audio
At Google I/O on 20 May 2025, Google introduced Veo 3, a video generation model that produces audio alongside video, together with the image model Imagen 4, the music model Lyria 2 and the AI filmmaking tool Flow.
Google opens the Jules coding agent in public beta
On 20 May 2025, Google opened Jules, its asynchronous coding agent, to everyone in a public beta; it runs on Gemini 2.5 Pro and works on users' GitHub repositories inside Google Cloud.