multimodal

Mistral AI has released Shieldstral-1.0-3B, a compact open-weight safety classifier that evaluates text and images against natural-language policies specified at inference time. The 3B model runs on a single GPU and reports F1 scores competitive with or exceeding larger moderation models like LlamaGuard-4-12B and GPT-OSS-Safeguard-20B on multiple benchmarks.

Mistral has released Shieldstral, a 3B open-weights safety classifier that reframes content moderation as a policy-adaptive question-answering task. The model claims to match or outperform guard models up to 7x its size on text safety and multimodal benchmarks, and runs on a single 16GB GPU.

Microsoft has released Mage-VL, a codec-native multimodal foundation model built on a from-scratch 4B-parameter visual encoder paired with Qwen3-4B-Instruct-2507. The model claims up to 3.5x inference speedup over uniform frame sampling and outperforms Qwen3-VL-4B on video and temporal-grounding benchmarks, according to Microsoft.

Unsloth has released GGUF quantizations of Kimi K3, a 2.8-trillion-parameter open-weight Mixture-of-Experts model from Moonshot AI with a 1-million-token context window and native vision support. The largest lossless quantization (Q8) weighs in at 1.56TB.

Microsoft Research AI Frontiers has released Fara1.5-27B, a 27-billion-parameter multimodal agent that completes web tasks by reading screenshots and emitting click/type/scroll commands. The model, fine-tuned from Qwen3.5-27B, ships under MIT license with a 262K-token context window and is designed to run alongside Microsoft's MagenticLite sandbox.

Alibaba has released Qwen 3.8, a 2.4 trillion parameter open-weight model that the company claims trails only Fable 5. The multimodal model processes images, videos, and documents, with a preview available through Alibaba's platforms at 10 percent of standard pricing.

Thinking Machines Lab released Inkling, a mixture-of-experts model with 975B total parameters and 41B active parameters, trained on 45 trillion tokens across text, images, audio, and video. The Apache 2.0-licensed model supports up to 1M context and debuts alongside Inkling-Small (276B-A12B), marking what observers call the strongest US-based open-weight release to date.

Alibaba unveiled Qwen3.8 at the World Artificial Intelligence Conference in Shanghai, claiming the 2.4 trillion parameter model ranks second only to Anthropic's Fable 5. The company provided no benchmark scores, model card, or independent verification to support the claim.

Thinking Machines Lab released Inkling, a Mixture-of-Experts transformer with 975B total parameters and 41B active parameters, trained on 45 trillion tokens of text, images, audio and video. The Apache-2.0 licensed model is designed as a base for fine-tuning rather than a frontier model.

Moonshot AI has released Kimi K3, an open-weight multimodal reasoning model with a 1-million token context window. The model is priced at $3 per 1M input tokens and $15 per 1M output tokens, available through OpenRouter.

Google has released Gemma 4 E2B for TPU, a variant of its open-source Gemma 4 model optimized to run natively on the Tensor G5 chip in Pixel 10 devices. The multimodal model enables completely offline AI chat, image recognition, and audio transcription on Pixel 10, 10 Pro, 10 Pro XL, and 10 Pro Fold.

NVIDIA released Audex-30B-A3B, a unified audio-text model built on the Nemotron-Cascade-2-30B-A3B backbone. The model handles audio understanding, speech recognition and translation, text-to-speech, audio generation, and speech-to-speech while supporting up to 1M token context length.

OpenAI has replaced ChatGPT's voice mode with GPT-Live, a new voice model that can delegate complex tasks to GPT-5.5 in the background. The previous voice mode was based on a GPT-4o era model with a 2024 knowledge cutoff.

Nex AGI has released Nex-N2-Mini, an open-source agentic mixture-of-experts model with a 262K-token context window. The model accepts text and image inputs and is priced at $0.025 per 1M input tokens and $0.10 per 1M output tokens.

Mistral AI has released Leanstral 1.5, an open-source 119B parameter mixture-of-experts model designed specifically for Lean 4 proof assistance. The model features 128 experts with 4 active per token (6.5B activated parameters), a 256k token context window, and multimodal input capabilities.

Portugal has released Amália, its first national AI model trained specifically for European Portuguese. Built on EuroLLM-9B with 9 billion parameters, the model is fully open-source with weights, datasets, and code published under an open license. The government has committed €5.5m in initial funding through 2027.

Google has released Gemini 3.1 Flash Lite Image, a text-to-image model that generates 1K resolution images in approximately 4 seconds — 2.7× faster than Gemini 3.1 Flash Image. The model is priced at $0.25 per 1M input tokens and $1.50 per 1M output tokens, with a 66K context window and knowledge cutoff of January 2025.

Proton has released Lumo 2.0, adding image recognition and generation, encrypted memory features, and enhanced web search to its privacy-focused AI assistant. The company claims Lumo 2.0 Max scored 240% higher than version 1.4 on the Artificial Analysis Intelligence Index, while maintaining zero-access encryption and no conversation logging.

AWS published a technical demonstration showing that pairing Amazon Nova 2 Lite with Anthropic's Claude Sonnet 4.6 reduces document processing costs by approximately two-thirds compared to single-model approaches. The two-stage pipeline processed 336 scanned yearbook pages at $0.0027 per page, producing 3,122 name-to-face associations with 93% scoring at or above 0.95 confidence.

Google has added a screen selection tool to Chrome 149's Gemini panel that allows users to capture text or images from their current tab for prompts. Separately, the company integrated computer use capabilities directly into the Gemini 3.5 Flash model API, replacing the standalone Gemini 2.5 Computer Use model.

Mistral AI released Mistral OCR 3, a document extraction model priced at $2 per 1,000 pages ($1 with Batch API discount). The model achieves a 74% overall win rate over its predecessor on forms, scanned documents, complex tables, and handwriting according to internal benchmarks.

Mistral has released Mistral 3, including Mistral Large 3—a sparse mixture-of-experts model with 41B active and 675B total parameters—and three Ministral 3 edge models (3B, 8B, 14B). All models are released under Apache 2.0 license with multimodal capabilities and are available today on multiple platforms.

Mistral AI has released a major update to Le Chat, adding a Deep Research agent that generates structured research reports, a new voice input model called Voxtral, and Projects for organizing conversations. The update also includes multilingual reasoning powered by Mistral's Magistral model.

Mistral AI has released Mistral OCR, an API for extracting text and images from documents at $1 per 1,000 pages (approximately $0.50 with batch inference). The company claims 94.89% overall accuracy on its internal test set, comparing favorably to GPT-4o (89.77%), Gemini 2.0 Flash (88.69%), and Azure OCR (89.52%).

Mistral AI has released major updates to le Chat, its free AI assistant, adding web search with citations, a canvas interface for collaborative editing, and image generation powered by Black Forest Labs' Flux Pro. The updates are powered by the new Pixtral Large multimodal model and are available free in beta.

Google has released Gemini 3.1 Flash Image, internally codenamed "Nano Banana 2," an image generation and editing model with a 131K context window. The model is priced at $0.50 per 1M input tokens and $3 per 1M output tokens.

Google has released Nano Banana Pro, an advanced image generation and editing model built on Gemini 3 Pro. The model supports 2K/4K output resolution, preserves identity across up to five subjects, and includes real-time Search grounding for context-rich visual synthesis.

NVIDIA released a quantized version of Google DeepMind's DiffusionGemma 26B A4B IT, a multimodal model with 25.2B total parameters (3.8B active) that processes text, image, and video inputs. The NVFP4-quantized model achieves generation speeds exceeding 1,100 tokens per second on NVIDIA H100 GPUs while supporting a 256K token context window.

MiniMax has released M3, a multimodal model with approximately 428 billion parameters and 23 billion activated parameters. The model supports a 1 million token context window and uses MiniMax Sparse Attention to achieve 9× prefill and 15× decode speedups compared to its predecessor M2.

Moonshot AI has released Kimi K2.7 Code, a 1 trillion parameter Mixture-of-Experts model designed for long-horizon coding tasks. The model features a 256K context window and reduces thinking token usage by approximately 30% compared to its predecessor K2.6.

Apple announced five third-generation foundation models at WWDC26, headlined by AFM 3 Core Advanced—a 20-billion-parameter sparse model that runs on-device by activating only 1-4 billion parameters at a time. For the first time, Apple extended Private Cloud Compute to third-party infrastructure, with AFM 3 Cloud Pro running on Nvidia GPUs in Google Cloud.

Google DeepMind released DiffusionGemma, a 26B parameter mixture-of-experts model that generates text using discrete diffusion instead of autoregression. The model processes blocks of 256 tokens in parallel, achieving generation speeds exceeding 1100 tokens per second on H100 GPUs in low-batch settings.

Anthropic has released Fable 5, making capabilities from its previously restricted Mythos model available to the public. The company claims Fable 5 beats GPT-5.5, Gemini 3.1 Pro, and its own Opus 4.8 in internal testing, with pricing set at $10 per million input tokens and $50 per million output tokens after a free trial period ending June 22.

Google is making AI Mode Pro's interactive visual generation capabilities available to all users for free through summer 2026, coinciding with the FIFA World Cup. The company is also adding real-time match tracking, tactical diagrams, and enhanced Gemini features for soccer fans.

Google announced Gemini 3.5 Live Translate, a speech-to-speech translation model supporting over 70 languages with continuous audio generation. The model rolls out today to Google Translate on Android and iOS, with Google Meet integration coming in private preview this month for select Workspace customers.

Google DeepMind has released Gemma 4 12B, a 12-billion parameter multimodal model that runs locally on laptops with 16GB of RAM. The model eliminates separate vision and audio encoders, processing raw inputs directly through its language model backbone under an Apache 2.0 license.

Google DeepMind has released quantization-aware training (QAT) optimized versions of its Gemma 4 model family in GGUF Q4_0 format. The QAT versions preserve similar quality to bfloat16 while dramatically reducing memory requirements, with models available across the entire Gemma 4 lineup: E2B, E4B, 12B, 26B A4B, and 31B.

Apple announced Siri AI at WWDC 2026, revealing a "deep collaboration with Google" that leverages Gemini models for its next-generation Apple Intelligence features. The new Siri includes personal context understanding, app actions, on-screen awareness, and conversational capabilities previously absent from the original Siri.

NVIDIA has released Nemotron 3.5 Content Safety, a 4B-parameter model built on Google Gemma 3 4B IT that provides multimodal safety classification across approximately 140 languages. The model includes a 128K context window, custom enterprise policy enforcement, auditable reasoning traces, and is releasing its training dataset.

Ideogram has released Ideogram 4, its first open-weight text-to-image model with 9.3 billion parameters. The model supports native 2K resolution, structured JSON prompting with bounding-box layout controls, and is available in nf4 and fp8 quantizations under a non-commercial license.

Google launched AI Edge Gallery for macOS, allowing Mac users to run Google's Gemma models locally. The platform ships with five Gemma models, including the newly released Gemma 4 12B—a 12-billion-parameter multimodal model that handles text, vision, and audio while running on consumer laptops with 16GB of RAM.

Ideogram has released Ideogram 4, a 9.3B parameter open-weight text-to-image model trained from scratch. The model features structured JSON prompting, native 2K resolution output, and ranks as the top open-weight model on Design Arena. Available in fp8 and nf4 quantizations under a non-commercial license.

Google DeepMind released Gemma 4, a family of open-weight multimodal models ranging from 2.3B to 30.7B parameters. The flagship 12B Unified model eliminates separate encoders, processing text, images, audio, and video directly through a single decoder-only transformer with up to 256K token context window.

Google DeepMind has released Gemma 4 12B Unified, an encoder-free multimodal model that processes text, images, and audio through a single decoder-only transformer. The model features 11.95 billion parameters, a 256K token context window, and achieves 77.2% on MMLU Pro and 72.0% on LiveCodeBench v6.

Alibaba's Qwen has released Qwen3.7 Plus, a multimodal model with a 1 million token context window. The model accepts text and image input with text output, priced at $0.40 per million input tokens and $1.60 per million output tokens through OpenRouter's API.

ByteDance released Bernini-R, an open-source video generation and editing model that combines an MLLM-based semantic planner with a DiT-based renderer. The model requires Hopper-class GPUs (H100/H800/H200) for optimal performance and supports multiple tasks including text-to-video, video editing, and reference-guided generation.

Microsoft announced MAI-Thinking-1, its first advanced reasoning AI model, at Build 2026. The company claims it's a medium-sized model matching leading models on key software engineering benchmarks, trained from scratch without distillation from third-party models.

NVIDIA released Cosmos3-Super-Text2Image, a 64-billion parameter text-to-image generation model as part of its Cosmos3 collection of omnimodal world models. The model uses a Mixture-of-Transformers architecture combining autoregressive and diffusion transformers, designed for Physical AI applications including robotics and autonomous vehicles.

NVIDIA released Cosmos 3, an omnimodal world foundation model platform for Physical AI spanning robotics, autonomous driving, and industrial environments. The flagship Cosmos3-Super variant contains 64 billion parameters and generates video, images, audio, and action commands from text, image, video, and action trajectory inputs using a Mixture-of-Transformers architecture.

NVIDIA released Cosmos3-Super, a 64-billion parameter omnimodal foundation model that generates video, images, audio, and action commands from combinations of text, image, video, and action trajectory inputs. The model, part of the Cosmos3 collection, targets Physical AI applications including robotics, autonomous vehicles, and industrial automation.