Kling: some farewells come without warning, but they can be recreated
Some farewells come without warning—but you can recreate them.
Some farewells come without warning—but you can recreate them.
OpenAI launched ChatGPT Images 2.5, which can turn ideas, sketches, and reference photos into more personalized, more refined images that stay closer to what users originally had in mind.
Google released Gemini 3.7 Flash, focused on coding and agents, priced at half the launch price of 3.6 Flash per million tokens; it also launched the speech-to-text model Gemini 3.5 Transcribe, aimed at voice agents, real-time captions and post-call analysis. The Gemini app has surpassed 1 billion monthly active users, and Gemini Omni 1.1 Flash is now available on Google Flow, AI Studio and other platforms.
Google DeepMind released Gemini Omni 1.1 Flash, offering generative video capabilities to developers through the Gemini API in Google AI Studio.
Why it matters: Gemini Omni 1.1 Flash opens scene extension, first/last frame interpolation and 4K upscaling to developers, giving them a basis to assess room for reworking video workflows.
NVIDIA released the Magpie TTS Multilingual open-weight model with 364M parameters, supporting 12 languages. This update adds Modern Standard Arabic, Korean and Brazilian Portuguese, and improves multilingual synthesis quality through updated training data.
Google released its new-generation music generation model Lyria 3.5 and launched it in Google Flow Music. The company says it improves musicality, lyrics and vocal quality, with richer and more natural melodic structure, stronger prompt adherence and structural awareness in lyric generation, improved vocal expressiveness and pronunciation, plus new controls for tempo and duration.
Google DeepMind has released two models: Nano Banana 2 Lite (gemini-3.1-flash-lite-image) and Gemini Omni Flash (gemini-omni-flash-preview).
Why it matters: The two models focus on speed and cost for image generation and conversational editing for video, respectively, allowing developers to evaluate integration into combined workflows.
Google has released the audio model Gemini 3.5 Live Translate, supporting near real-time speech-to-speech translation across 70+ languages, with automatic language detection and preservation of the speaker's tone, rhythm, and pitch.
Why it matters: Gemini 3.5 Live Translate reduces simultaneous interpretation latency to a few seconds and supports 70+ languages, enabling developers to evaluate real-time dubbing and meeting translation solutions.
Google DeepMind has released Gemini Omni Flash, the first model in the Gemini Omni family, which can combine images, audio, video, and text as inputs to generate high-quality video and supports multi-turn editing through natural language conversation.
Why it matters: Gemini's first video generation model, Omni Flash, is now live, supporting multimodal inputs and conversational multi-turn editing, allowing evaluation against existing video production workflows.
OpenAI has launched a new real-time voice model in its API, capable of reasoning, translation, and speech transcription, to support more natural and intelligent voice experiences.
NVIDIA has launched the Nemotron 3 Nano Omni omnimodal understanding model, featuring a Nemotron 3 hybrid Mamba-Transformer MoE backbone paired with the C-RADIOv4-H visual encoder and the Parakeet-TDT-0.6B-v2 audio encoder, supporting 100+ page documents, long audio and video understanding, and agentic computer operation.
OpenAI has launched ChatGPT Images 2.0, improving text rendering, multilingual support, and visual reasoning capabilities. This version was released in 2026; new features in Images 2.5 can be viewed separately.
Google has released Gemini 3.1 Flash TTS, calling it the most natural and expressive voice model to date, achieving 1,211 Elo points on the Artificial Analysis TTS leaderboard.
Why it matters: Gemini 3.1 Flash TTS introduces audio tags and multi-speaker dialogue, enabling developers to evaluate improvements in controllability for voice generation.
Overworld releases the real-time video world model Waypoint-1.5, which can generate real-time environments at up to 720p and 60 FPS on desktop hardware such as RTX 3090 to 5090, and adds a 360p tier for broader consumer-grade hardware, with Apple Silicon Mac support to follow.
Why it matters: Waypoint-1.5 brings real-time world models down to consumer-grade GPUs, with 720p and 360p tiers letting creators evaluate the feasibility of local interactive generation.
Google releases Gemini 3.1 Flash Live, calling it the highest-quality audio and voice model to date, with higher accuracy and lower latency.
Why it matters: Gemini 3.1 Flash Live provides comparable benchmark data on latency, long-range reasoning, and multilingual performance for voice interactions, enabling developers to evaluate the deployment of voice agents.
Google DeepMind releases Lyria 3 Pro, which can generate tracks up to 3 minutes long and supports prompting by structural elements such as intro, verse, chorus, and bridge. This version is now available on Vertex AI public preview, Google AI Studio, and the Gemini API, and is integrated into Google Vids, the Gemini app, and ProducerAI; outputs from both Lyria 3 and Lyria 3 Pro are embedded with SynthID watermarks.
Why it matters: Lyria 3 Pro extends single-track length to 3 minutes and supports section-level prompting, allowing creators to assess whether it can be used for finished film scoring.
Addressing the new safety challenges posed by the Sora 2 video model and the Sora social creation platform, OpenAI treats safety as a foundational design principle, supported by specific protective measures.
Google DeepMind releases its latest image generation model, Nano Banana 2 (Gemini 3.1 Flash Image), bringing Nano Banana Pro's advanced world knowledge, quality, and reasoning capabilities to Flash speed.
Why it matters: Nano Banana 2 brings Pro-level world knowledge and text rendering down to Flash speed, with multi-platform availability and pricing information, enabling assessment of whether to switch image workflows.
Google DeepMind launches its latest music generation model, Lyria 3, in beta within the Gemini app, allowing users to generate 30-second songs from text or images, with lyrics automatically generated by the model and control over style, vocals, and rhythm.
Why it matters: Lyria 3 opens up text- and image-based 30-second song generation within the Gemini app and adds audio SynthID verification, offering insight into the productization path for music generation.
Overworld releases the real-time interactive video diffusion model Waypoint-1, controllable via text, mouse, and keyboard. Weights are now available on Hugging Face Hub, including Waypoint-1-Small, with Waypoint-1-Medium coming soon.
Why it matters: The real-time interactive video diffusion model is open-sourced, allowing evaluation of its real-world performance in interactive world generation on consumer-grade hardware.
Google DeepMind introduces D4RT (Dynamic 4D Reconstruction and Tracking), an AI model that unifies 4D scene reconstruction and tracking. It employs an encoder-decoder Transformer architecture with a query mechanism, enabling simultaneous point tracking, point cloud reconstruction, and camera pose estimation.
Google DeepMind updates Veo 3.1 Ingredients to Video, making reference-image-based video generation more expressive and adding native 9:16 vertical output for the first time, targeting mobile short-form content such as YouTube Shorts.
Why it matters: The Veo 3.1 Ingredients to Video update brings native vertical output and 1080p/4K upscaling, enabling creators to evaluate short-drama production workflows.
NVIDIA releases the open-source reasoning vision-language model Cosmos Reason 2, which ranks first among open-source visual understanding models on the Physical AI Bench and Physical Reasoning leaderboards.
OpenAI launches ChatGPT Images and GPT-Image-1.5. The former provides a faster image generation experience, while the latter is available via API.
Google releases the upgraded Gemini 2.5 Flash Native Audio, designed for real-time voice agents. It improves complex workflow handling, instruction following, and multi-turn dialogue capabilities, leading on ComplexFuncBench Audio with 71.5%, and raising developer instruction-following rates from 84% to 90%.
Why it matters: Gemini 2.5 Flash Native Audio updates function calling, instruction following, and multi-turn dialogue capabilities, and adds real-time voice translation, enabling evaluation of usability in voice agent and translation scenarios.
Hugging Face Diffusers announces support for FLUX.2, providing invocation code for Flux2Pipeline and Flux2Transformer2DModel. The example uses the diffusers/FLUX.2-dev-bnb-4bit repository with the Mistral3 text encoder and enables enable_model_cpu_offload.
Why it matters: Diffusers integrates FLUX.2 and provides a 4-bit quantized loading example, enabling creators to evaluate the barrier to local deployment.
Google DeepMind releases Nano Banana Pro (Gemini 3 Pro Image), a high-fidelity image generation and editing model built on Gemini 3 Pro, now available to developers as a paid preview on Google AI Studio and Vertex AI.
Why it matters: Gemini 3 Pro Image is available to developers as a paid preview, supporting 2K/4K output and multi-image composition, enabling evaluation of existing image generation workflows.
Google DeepMind releases Nano Banana Pro (Gemini 3 Pro Image), built on Gemini 3 Pro, featuring more accurate in-image text rendering, multilingual generation, and real-time information visualization.
Why it matters: The Gemini 3 Pro-based image model introduces new capabilities in text rendering and multi-image consistency, enabling evaluation of image generation workflows.
OpenAI releases the next-generation video and audio generation model Sora 2 and publishes its system card. Compared to Sora, the new model shows improvements in physical accuracy, realism, synchronized audio, controllability, and style range; the official statement notes that these capabilities were difficult for previous video models to achieve.
Why it matters: OpenAI publishes the system card for Sora 2, listing capability changes such as physical accuracy and synchronized audio, allowing comparison with the previous generation to assess the upgrade magnitude.
OpenAI releases the video generation model Sora 2, which can generate dialogue and sound effects synchronized with the visuals. The Sora product that this model belongs to is currently no longer available.
OpenAI releases Sora 2 and the Sora app, calling it the most advanced video model and a new social creation platform, and in response to the new safety challenges it brings, builds safety into the foundation and takes specific protective measures.
Why it matters: OpenAI simultaneously explains its safety design approach when releasing Sora 2 and the Sora app, providing insight into the official framework for addressing safety challenges in video models.
OpenAI announces that its latest image generation model is now available in the API as gpt-image-1, enabling developers and businesses to directly integrate professional-grade, customizable visual content into their own tools and platforms.
Why it matters: OpenAI has integrated its latest image generation model into the API as gpt-image-1, allowing developers to evaluate image generation integration for their own tools.
The Language Technology Lab releases Visual Salamandra, extending the capabilities of the Salamandra large model to images and video. Based on a 7-billion-parameter foundation model, it integrates the Google SigLIP-So400m encoder with a 2-layer MLP projector, employing a four-stage training process and a late-fusion architecture.
OpenAI releases native image generation capabilities in GPT-4o. The company states that it is stronger in text rendering and instruction following, and points to ChatGPT Images 2.5.
Why it matters: OpenAI adds native image generation to GPT-4o, with text rendering and instruction following being the two key improvements in this update.
OpenAI releases a GPT-4o system card addendum introducing 4o image generation. This capability is stronger than the previous DALL·E 3 series models, capable of generating photorealistic images and accepting image inputs for transformation.
Why it matters: OpenAI outlines the capability changes of 4o image generation compared to DALL·E 3 in the system card addendum, providing insight into its positioning for image input and photorealistic output.
OpenAI launches next-generation audio models in the API. Developers can now instruct text-to-speech models to speak in specific ways, such as "speak like an empathetic customer service agent," bringing new customization capabilities to voice agents.
Hugging Face releases the SmolVLM2 series of video language models, available in three parameter sizes: 2.2B, 500M, and 256M. The 500M and 256M variants are billed as the smallest video language models to date.
OpenAI's video generation model Sora is now available at sora.com. Users can generate videos up to 1080p resolution and up to 20 seconds in length, supporting widescreen, portrait, and square aspect ratios. Users can import their own assets for extension, remixing, and blending, or generate entirely new content from text.
Why it matters: Sora is now officially available, and creators can learn about its basic generation specifications including resolution, duration, and aspect ratio.
Filmmaker Lyndon Barrois shares how he uses Sora as a storytelling tool.
OpenAI publishes the system card for its video generation model Sora. Sora can accept text, image, and video inputs to generate new videos. Its design draws on lessons from DALL-E and GPT models, aiming to provide richer tools for storytelling and creative expression.