MiniCPM-o 4.5
PulseAugur coverage of MiniCPM-o 4.5 — every cluster mentioning MiniCPM-o 4.5 across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
New benchmark DuplexLedger reveals memory gaps in full-duplex AI agents
Researchers have introduced DuplexLedger, a new benchmark designed to evaluate the transactional memory capabilities of full-duplex speech agents. The benchmark comprises 1,094 sessions across four scenarios, providing …
-
OmniTrace framework enables source attribution in omni-modal LLMs
Researchers have developed OmniTrace, a novel framework designed to attribute generated text in omni-modal large language models to specific parts of the input data. This method operates during the generation pass, util…
-
LEAP framework enhances audio-video Q&A for long recordings
Researchers have developed LEAP, a novel framework designed to improve audio-visual question answering for hour-long recordings. LEAP addresses the context length limitations by dividing recordings into blocks and using…
-
Speech models achieve 64% KV cache reduction with new text compression
Researchers have developed a novel method for compressing key-value (KV) states in full-duplex speech models, significantly reducing memory usage during long-running interactions. This technique, termed acoustic-to-text…
-
New 'Latent Bridge' enhances real-time AI agents for gaming
Researchers have developed a novel 'Latent Bridge' technique to improve real-time AI agents for tasks like gaming. This method couples a slow, reasoning-capable VLM with a fast, reactive VLM by projecting the slow model…
-
Ex-OpenAI researcher's AI vision echoes Chinese firm's prior release
Former OpenAI researcher Lilian Weng's new venture, Thinking Machines Lab (TML), has unveiled a vision for full-duplex, real-time conversational AI. This concept closely mirrors the capabilities demonstrated by China's …
-
Consumer-grade graphics cards can quickly get started! MiniCPM-o 4.5 from Mianbi Intelligent releases technical report
MiniCPM-o 4.5 is a new 9B parameter omni-modal large language model designed for real-time, full-duplex interaction. It can simultaneously process and generate audio, video, and text, enabling proactive behaviors and co…