MiniMax's Olive Song: AI Agents Need Million-Token Context, Not Just Bigger Text Models
MiniMax CTO Olive Song argues the next generation of AI agents can't rely on text-only models with vision bolted on afterward. She says true multimodal reasoning at massive context scale is essential. Source: finance.biggo.com
Importance:Newsefficiency
AI 'recalibration' technique lets small models perform like GPT-4V
A new recalibration method reportedly allows smaller AI models to reach performance levels comparable to GPT-4V. This could make advanced multimodal capabilities more accessible without the computational cost of massive models. Source: quantumzeitgeist.com
Importance:LaunchGemini video generation
Google unveils Gemini Omni 1.1 Flash, generating 4K AI videos up to 40 seconds long
Google has launched a new version of its multimodal model capable of producing longer AI-generated videos in 4K resolution. The update pushes video length limits well beyond earlier Gemini releases. Source: neowin.net
Importance:Newsmultimodal AI
What exactly is multimodal AI, and how does it change things when one model can read, see and hear?
Multimodal AI refers to systems that combine text, image, and audio understanding within a single model instead of relying on separate tools. This approach enables more natural interactions and broader use cases, from analyzing photos to processing spoken language. Source: sqmagazine.co.uk
Importance:Researchmultimodal generation
STARFlow2 combines language models and normalizing flows for multimodal AI generation
Researchers have introduced STARFlow2, a new approach that merges large language models with normalizing flow techniques to enable unified generation across different data types. The method targets more consistent and flexible multimodal content creation. Source: machinelearning.apple.com
Importance:Researchmultimodal models
STARFlow2 combines language models and normalizing flows for multimodal generation
STARFlow2 is a new architecture that merges language models with normalizing flows to enable unified generation across multiple data modalities. The approach seeks to bridge text and other content types within a single generative framework. Source: machinelearning.apple.com
Importance:Launchmultimodal AI models
Black Forest Labs Unveils FLUX 3 for Multimodal Video Generation
FLUX 3 is a new foundation model from Black Forest Labs built to jointly handle images, video, and audio, with plans to eventually add text-based "action prediction" capabilities. Source: dynamicbusiness.com
Importance:Launchmultimodal AI video generation
Seedance 2.0 Adds Multimodal AI Video for Maps, Urban Planning and Location Content
Turning spatial data into video used to require drone footage, 3D rendering pipelines, or motion graphics teams. ByteDance's revamped AI video model now aims to automate that process for geospatial visualization and location-based content. Source: gisuser.com
Importance:Launchmultimodal models
DeepSeek unveils multimodal model rivaling Opus 4.8
The new V4 Flash Vision Exp model is initially available only through the Chinese startup's paid developer platform. DeepSeek has hinted a free version could follow later. Source: siliconangle.com
Importance:ResearchBiomedical foundation models
New AI Model Explains Its Medical Image Diagnoses Using Concepts
Researchers built a biomedical foundation model that combines vision and language pretraining with explicit medical concepts, aiming to make AI diagnoses more interpretable for doctors. The approach addresses a key limitation of current multimodal medical AI systems, which often lack transparency. Source: nature.com
Importance:Launchvideo generation tool
Seedance 2.5 Launches on AI Inspo With 30-Second Clips and Native Audio
AI Inspo has added Seedance 2.5 to its creative platform, enabling generation of 30-second AI videos from multimodal inputs with built-in audio. The update expands the tool's capabilities for longer, more complex AI-generated content. Source: somdnews.com
Importance:LaunchAI video tools
Seedance 2.5 Arrives on AI Inspo, Enabling 30-Second AI Videos with Multimodal Input and Native Audio
AI Inspo has added Seedance 2.5 to its creative platform, allowing users to generate AI videos up to 30 seconds long. The model supports multimodal inputs and comes with built-in audio generation. Source: streetinsider.com
Multimodal LLM Tracks Disease Progression in Oral Lichen Planus Study
A longitudinal study tested a multimodal LLM's ability to classify disease trajectories and stratify risk in patients with oral lichen planus. The research assessed how accurately the model could follow changes over time compared to traditional diagnostic methods. Source: nature.com
Importance:NewsMiniMax model performance
MiniMax M3 Scores 55 on Index, Beats GPT-5.5 in Benchmarks
MiniMax M3 has launched with a 1-million-token context window, native multimodal capabilities, and pricing starting at $0.30 per million tokens. The model reportedly outperforms GPT-5.5 on the SWE-Bench Pro benchmark. Source: tech-insider.org
Importance:LaunchAI video generation tool
Pippit Debuts Seedance 2.5: 4K AI Video Up to 30 Seconds With Finer Creative Controls
Pippit's new Seedance 2.5 update brings 4K video generation up to 30 seconds long, along with second-level timestamp precision and support for multiple input types. New features include multimodal references, audio-only prompts, 3D wireframe guidance, and improved tools for building narrative sequences. Source: manilatimes.net
Importance:ResearchAI alignment
Study Digs Into How Preference Alignment Shapes Multimodal LLMs
A new comprehensive study examines preference alignment, a technique widely used to boost LLM performance, and explores how it specifically affects multimodal models. The research looks at both the benefits and limitations of applying this method beyond text-only systems. Source: google.com
Importance:NewsAI Video Models
MiniMax H3: What You Need to Know About the New AI Video Model
MiniMax H3 is an open-weight AI video generator built on multimodal reasoning, contextual regeneration, and an advanced architecture. Source: ca.news.yahoo.com
Importance:ResearchLLM alignment
A Closer Look at Alignment in Multimodal LLMs
Preference alignment has become key to improving performance of large language models, but its effects in multimodal settings remain less understood. A new comprehensive study examines how alignment techniques play out across different modalities. Source: machinelearning.apple.com
Importance:NewsAI model comparison
FLUX 3 Combines Image, Video and More in One Model, Outpacing Seedance 2.0 and Grok
Black Forest Labs' new FLUX 3 model aims to replace the traditional patchwork of separate tools for image generation, video synthesis and other tasks with a single multimodal system. According to early comparisons, it outperforms rivals like Seedance 2.0 and Grok in several benchmarks. Source: sitepoint.com
Importance:Newsmodel releases
MiniMax Launches H3 Video Model Amid Copyright Controversy
MiniMax has released H3, a multimodal video generation model that reportedly tops independent benchmarks for video editing among all tracked models. However, the launch comes as the company faces a copyright lawsuit tied to its video generation technology. Source: techtimes.com