Top Stories
All Sections
Finance
About
13 results for "multimodal"
Meta's Muse Ember 1.0 enters the high-volume agent model tier at $0.10 per million input tokens, with native support for text, images, audio, and video and a 1-million-token context window.
Mistral released the full weights of Le Chonk, its trillion-parameter multimodal model, under Apache 2.0 after RL fine-tuning boosted its Terminal-Bench 4.0 score from 28% to 41% and cut output tokens by 30%.
Tencent released Hunyuan Hy4, a 770-billion-parameter open-weight model with vision understanding, computer-use capabilities, and a 1M-token context window — the largest openly released Chinese AI model to date.
A proof-of-concept study demonstrates that AI foundation models trained on wearable sensor data can predict migraines a full 24 hours before onset — the first multimodal wearable AI demonstration for migraine forecasting.
Mistral has open-sourced the weights for Mistral Large 4 — nicknamed Le Chonk — making it the largest open-weight multimodal model available. The 1.05 trillion-parameter release includes day-one vLLM support for self-hosted deployment.
Liquid AI has launched LFM3, a model family that processes text, images, and audio within a single shared architecture — now available locally via Ollama on all major platforms.
Mistral AI released a preview of Le Chonk, a 1.05 trillion-parameter multimodal model — Europe's first entry into frontier-scale AI, with open weights promised by October 31.
Google released Gemma 4 12B under Apache 2.0 on June 3, adding audio, vision, and a 256K context window to an openly licensable model — making frontier multimodal AI accessible without API fees for the first time.
Mayo Clinic and Microsoft are jointly training a domain-specific frontier AI model on Mayo's clinical datasets — targeting multimodal reasoning across imaging, labs, and notes for earlier diagnosis and personalized treatment.
NVIDIA released Nemotron 3 Nano Omni, a 30B-parameter open model delivering 9x the throughput of comparable omni models — a direct play to own the agentic AI stack, not just the chips.
Meta released Muse Spark, its first major AI model in over a year, deploying multimodal reasoning across WhatsApp, Instagram, Facebook, Messenger, and Meta AI glasses for three billion users.
Meta debuted Muse Spark on April 8 — a proprietary multimodal model from its new Superintelligence Labs, marking a sharp break from the open-source Llama playbook.
Meta Superintelligence Labs launched Muse Spark, a natively multimodal model with tool use and multi-agent capabilities, backed by $135B in 2026 AI capex.