A "free" multi-modal "embedding" model (already) included with your LLM: vLLM plugin + upstream PR
TLDR: Text/image (simili-)embeddings from the LLM you already have loaded :-)
Proof1 community thread
Key points
- Disclaimer: I work for NuMind, the company behind this plugin.
- I've been working on making it easier to obtain embeddings from a LLM (preferably multi-modal) that is already being served by vLLM.
- The idea is simple (and not mine as we'll discuss): the model already computes internal representations (1 hidden state per decoder layer) of your prompt to predict the first output token.
- We expose the last hidden state vector through the chat API (compatible with the openai-api and openai client even though it's not standard)
Sources (1)
- [1]A "free" multi-modal "embedding" model (already) included with your LLM: vLLM plugin + upstream PRr/LocalLLaMA (top, daily) · Oct 12, 12:59 PM
TLDR: Text/image (simili-)embeddings from the LLM you already have loaded :-)
Disclaimer: I work for NuMind, the company behind this plugin.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 12, 2026StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions
- Oct 11, 2026Taming the AI Hardware Cambrian Explosion — Abdul Dakkak, Modular
- Oct 10, 2026When the Safety Test Became the Threat: The Machine That Found Its Own Way Out
- Oct 10, 2026Prior-Fitted and Frozen Scorers for Multimodal Contrastive Learning
- Oct 10, 2026Bidirectional Adaptive Transformers for Multimodal Anomaly Detection
- Oct 9, 2026Microsoft's Decision-1 model enters the fast-growing AI decision model race
