🤖 AI Brief

Five labs, five minds: building a multi-model finance drama on small models

The article discusses a collaborative project involving five labs to build a multi-model finance drama using small models.

Thousand Token Wood: shipping a multi-agent economy on a 3B model

A multi-agent economy is deployed on a 3B parameter model, demonstrating agentic AI capabilities at small scale.

Model Release: Thousand Token Wood

Nemotron 3.5 Content Safety: Customizable Multimodal Safety for Global Enterprise AI

NVIDIA releases Nemotron 3.5 Content Safety, a customizable multimodal safety model designed for enterprise AI applications across global markets.

Model Release: Nemotron 3.5 Content Safety (NVIDIA)

EVA-Bench Data 2.0: 3 Domains, 121 Tools, 213 Scenarios

EVA-Bench Data 2.0 expands to cover 3 domains, 121 tools, and 213 scenarios for evaluating AI agent performance.

Direct Preference Optimization Beyond Chatbots

No summary available.

Holo3.1: Fast & Local Computer Use Agents

Holo3.1 is a fast, locally-run computer use agent model released on Hugging Face.

Model Release: Holo3.1 (Holo)

Introducing Mellum2: A 12B Mixture-of-Experts Model by JetBrains

JetBrains introduces Mellum2, a 12B parameter mixture-of-experts model, available on Hugging Face.

Model Release: Mellum2 (JetBrains)

Beyond LLMs: Why Scalable Enterprise AI Adoption Depends on Agent Logic

A blog post argues that scalable enterprise AI adoption depends on agent logic beyond LLMs, focusing on reasoning and action orchestration.

ITBench-AA: Frontier Models Score Below 50% on the First Benchmark for Agentic Enterprise IT Tasks — by Artificial Analysis and IBM

Frontier models score below 50% on ITBench-AA, the first benchmark for agentic enterprise IT tasks, created by Artificial Analysis and IBM.

Harness, Scaffold, and the AI Agent Terms Worth Getting Right

A blog post discussing terminology for AI agent systems, focusing on 'harness' and 'scaffold' concepts.