OpenAI today announced that GPT-Realtime voice model is now available, supporting image input, marking the Realtime API’s move from beta to general availability for production voice agents. The release positions GPT-Realtime as a low-latency, speech-to-speech model that can run two-way voice conversations while also grounding responses in images supplied during a session. OpenAI describes gpt-realtime […]
August 28, 2025 — xAI today introduced Grok Code Fast 1, a coding-focused variant in the Grok family designed to prioritize low latency and low cost for IDE integrations, agentic coding workflows, and large-codebase reasoning.The model is appearing as an opt-in public preview inside GitHub Copilot (VS Code) and is also available through xAI’s API […]
Google lately unveiled Gemini 2.5 Flash Image — a native, high-performance image generation and editing model that brings real-time, conversational image creation and precise, multi-step editing directly into the Gemini product family and developer tools. The release, described by Google as a “state-of-the-art” update to Gemini’s multimodal stack, is positioned for both consumer creativity and […]
ByteDance’s Seed team has released Seed-OSS, a family of open-source large language models led by Seed-OSS-36B, a 36-billion-parameter model that supports exceptionally long input windows and is being distributed under an Apache-2.0 license. The code and model cards were published on GitHub and Hugging Face on Aug. 20, 2025, and multiple variants — including a […]
Grok Imagine 0.1 is xAI’s new built-in image-and-video generator inside the Grok/X ecosystem. It lets users create images from text or voice prompts, and convert images into short videos with auto-generated sound. The tool launched as an early “0.1” release (explicitly described by Elon Musk as a beta) and has drawn both praise for speed […]
Midjourney’s HD video mode goes live — higher fidelity, higher cost, wider availability: Midjourney officially rolled out an HD video mode for its newly introduced video tools, opening higher-resolution AI video rendering to paying professional users. The addition upgrades Midjourney’s image-to-video workflow with a higher-pixel option that the company says targets creators who need crisper, […]
In a move that underlines how quickly generative AI is moving beyond text and images, Google DeepMind today unveiled Genie 3, a general-purpose “world model” capable of turning simple text or image prompts into navigable, interactive 3D environments that run in real time. The system represents a leap from previous generative-video and world-model experiments: Genie […]
OpenAI has announced the release of GPT-OSS, a family of two open-weight language models—gpt-oss-120b and gpt-oss-20b—under the permissive Apache 2.0 license, marking its first major open-weight offering since GPT-2. The announcement, published on August 5, 2025, emphasizes that these models deliver state-of-the-art reasoning performance at a fraction of the cost associated with proprietary alternatives, and […]
On August 5, 2025, Anthropic publicly released Claude Opus 4.1, a significant refinement of its flagship Opus 4 model family, aimed at advancing agentic tasks, real-world software engineering, and complex reasoning. This incremental update, which builds on the May debut of Claude Opus 4, delivers higher accuracy on coding benchmarks, extended context handling, and maintains […]
On August 4, 2025, Alibaba’s Qwen team officially launched Qwen-Image, a 20 billion-parameter multimodal diffusion transformer (MMDiT) foundation model designed to deliver unprecedented fidelity in text-to-image synthesis and precision image editing. This release marks Alibaba’s bold entry into the open-source image generation arena, positioning Qwen-Image as a direct challenger to proprietary systems like OpenAI’s GPT-4o, […]
