AI Models
Image, video and 3D generation models and the tools built on them.
Gemini 4 Carbon Launches as Google Shares 3D Generation Examples
Google made Gemini 4 Carbon available to developers, businesses, scientists and consumers on 10 November. The company also shared examples of the model creating interactive 3D scenes from plans.
Google and Adobe Expand Partnership: Acrobat and AI Integrations
Google and Adobe announced an expanded partnership on 10 November 2026 at Adobe MAX. The plan covers access to Acrobat from Drive and Gmail, Adobe tools in Gemini, and new editing possibilities for Photoshop.
AI imageAdobe MAX 2026: Native MCP and New AI Tools for Creative Cloud
At its MAX event in Miami, Adobe introduced native MCP integrations that will open Creative Cloud apps to third-party AI agents. It also announced new tools for Photoshop, Premiere, Lightroom and Firefly.
AI imageArtStation Expands AI Disclosure and Default Content Filter
ArtStation has introduced new disclosure options for generative AI use in projects, along with a default AI filter. Users can also adjust how strictly suspected content is filtered.
AI imageMeta introduces Muse Ember: Focused on repetitive agent tasks
Meta has made Muse Ember available for repetitive agent tasks such as tool calling, routing, and document processing. The model is accessible via the API and Muse Code, with pricing based on input and output tokens.
AI imageClaude Fable 5.2 and Mythos 5.2 Announced: Fable Available Today
Anthropic has announced Claude Fable 5.2 and Mythos 5.2, saying they bring improvements in coding, reasoning, and vision. Fable is available today, while access to Mythos will initially be limited to approved organizations.
AI imageNVIDIA–South Korea partnership expands to sovereign AI
NVIDIA and South Korea’s Ministry of Science and ICT are expanding their collaboration on the country’s sovereign AI ecosystem. The plan covers chips, data center infrastructure, training, and model inference.
AI imageNVIDIA to present certification roadmap: webinar on November 19
NVIDIA will present its updated certification roadmap and two new Certified Associate certifications at a webinar on November 19, 2026. Attendees will reportedly receive a 50% discount code for an exam.
AI imageNIST’s INSIGHT tool maps building interiors in 3D for emergency response
NIST is testing INSIGHT, a system that converts building scans into structured 3D data for emergency responders. The system aims to verify automated detections through human review.
AI imageNVIDIA Cosmos 3 adds four models for interactive simulation
On 5 November, NVIDIA added four open-weight models to the Cosmos 3 family for driving simulation and robotics. According to the company’s announcement, the models can generate camera, LiDAR and video data.
AI imageERNIE 5.5 Flash Base launches: An open model with 280 billion parameters
Baidu released ERNIE 5.5 Flash Base, an open foundation model with 280 billion parameters, on 5 November. The model is available without fine-tuning or post-training, and its weights are available on Hugging Face in FP8 and BF16 formats.
AI imageGemini Omni 1.2 Flash launches: 15-second video and multi-turn editing
Google DeepMind announced Gemini Omni 1.2 Flash on November 5. The model offers video generation up to 15 seconds and conversational editing, while lowering the price of 720p generation to $0.07 per second.
AI imageOpenAI Images 2.6 launches with GIFs, SVGs and faster image generation
OpenAI announced Images 2.6 for ChatGPT, Codex and the API on 4 November. The new version supports GIF generation, SVG export through the API and image editing with annotations, while the company reports generation speeds 50% faster than the previous version.
AI imageGPT-6 Astra edited model weights without retraining
An MIT study covered in The Decoder’s 4 November report describes experiments in which GPT-6 Astra directly examined and changed model weights. Qwen3.5-0.8B’s MATH-500 score rose from 7.9% to 48.1%.
AI imageSambaNova to Pair SN50 RDUs with NVIDIA GPUs in Supermicro Systems
SambaNova announced a collaboration with Supermicro to build AI inference systems combining SN50 RDUs and NVIDIA GPUs. The initial plan calls for 10,000 SN50 chips.
AI imageOpenAI Brings AI to CAD and Simulation Workflows in Engineering
OpenAI explained how its models are being used in CAD design, manufacturing drawings and simulation preparation. The company is running early experiments aimed at reducing repetitive work in fields including architecture and BIM.
TraVisionLM: An 875-Million-Parameter Model That Understands Turkish Images
Developed by Ucsahin, TraVisionLM can generate short captions, detailed descriptions and basic visual question-answering responses from an image and a Turkish instruction. The model has 875 million parameters and is licensed under Apache 2.0.
AI imagePika H3 Flex comes to ComfyUI: 17% faster in a seven-clip test
Pika released the weights for its MiniMax H3-based H3 Flex model for free on 30 October. In a seven-clip test on an RTX 3060, the model finished 17% faster than FastH3 and preserved reference images more faithfully.
AI imageGPT Image 2.5 Takes First Place in the November Model League
GPT Image 2.5 took first place in the November 2026 Model League with a score of 89.9. Three different models stood out across five architectural prompts, while speed and per-image cost helped set the results apart.
AI imageApple Intelligence in China: Qwen support arrives with iOS 27.3
Apple is integrating Alibaba’s Qwen models into Apple Intelligence in China. The features will be tested in the iOS 27.3 developer beta on 4 November, with wider availability planned for December.
OpenAI aims to implement shared robots.txt signal in coming months
OpenAI announced it will follow a shared signal designed to let websites communicate their AI training preferences to multiple providers through robots.txt. The company aims to implement the signal in the coming months.
AI imageBrainChip AKD1500 Kit Brings On-Device AI to Raspberry Pi 5
BrainChip has announced a development kit combining the Raspberry Pi 5 with its AKD1500 M.2 card. The kit aims to simplify the development of always-on AI applications on-device and the transition from prototype to production.
AI imageMistral Large 4 weights released, enabling on-premises deployment for organizations
Mistral made the weights for its trillion-parameter Mistral Large 4 model available for download on October 29. Organizations can obtain the model through Hugging Face or the API and deploy it on their own systems.
NVIDIA RTX Spark Update Makes GPU Resources Available to Local AI
NVIDIA’s 28 October RTX Spark update lets users allocate GPU resources between foreground apps and background agents. The package also includes local speech and image-generation models.
About this category
In the AI Models category, we cover developments in AI models that generate images, video, and 3D content, with a focus on architecture and visualization. Here you’ll find new releases, pricing and availability updates, and features that are genuinely useful for architectural rendering from models such as GPT Image, Nano Banana, Seedream, FLUX, Veo, and Kling.
We compile each story from official announcements and reliable industry sources, and consider what it means for architects and visualization artists. This category also features our monthly Model League results, comparing models with our own test set, as well as workflows for rapid AI-assisted concept development, material variations, and turning images into video.
AI models change quickly, so we clearly state each model’s version, release date, and access terms (API, subscription, or open weights). We also note practical details such as commercial usage rights, availability in Türkiye, and Turkish-language prompt support.