Media Summary: First calibration run of a cross-episode memory ARC AGI 3 launched a few weeks before this talk with every task human solvable and frontier models under 1%. That gap is the ... NVIDIA Blackwell delivered leading results in the

Agentperf Trajectory Replay Benchmarking Agents - Detailed Analysis & Overview

First calibration run of a cross-episode memory ARC AGI 3 launched a few weeks before this talk with every task human solvable and frontier models under 1%. That gap is the ... NVIDIA Blackwell delivered leading results in the In this AI Research Roundup episode, Alex discusses the paper: 'LocalSearchBench: Join us as we cover features of Dynamo and walk you through a hands-on demo. See how Dynamo accelerates inference for ...

Photo Gallery

AgentPerf — Trajectory-replay benchmarking (agents per megawatt)
AgentPerf - First Infrastructure Benchmark for Agents
Compiled Trajectory Replay — how PreAct makes agents 8.5–13× faster
AgenticDataBench: A Comprehensive Benchmark for Data Agents
Tier 1 Calibration Run — World Model Capability Benchmark (Braitenberg Agent)
The Art & Science of Benchmarking Agents — Vincent Chen, Snorkel AI
Agent Arena: The Benchmark That Shows How AI Agents Actually Fail
Agents' Last Exam: Benchmarking AI Agents on 1K+ Real-World Economic Tasks
NVIDIA Blackwell Leads AgentPerf: 20x Agentic AI Infrastructure Performance Explained
We built our own agent benchmark (and you should too)
LocalSearchBench: Benchmark for Agentic Local Search
Evaluating AI Agents: Why Memory-Wipe Benchmarks Hide Real Learning
View Detailed Profile
AgentPerf — Trajectory-replay benchmarking (agents per megawatt)

AgentPerf — Trajectory-replay benchmarking (agents per megawatt)

What is

AgentPerf - First Infrastructure Benchmark for Agents

AgentPerf - First Infrastructure Benchmark for Agents

Artificial Analysis released

Compiled Trajectory Replay — how PreAct makes agents 8.5–13× faster

Compiled Trajectory Replay — how PreAct makes agents 8.5–13× faster

Compiled

AgenticDataBench: A Comprehensive Benchmark for Data Agents

AgenticDataBench: A Comprehensive Benchmark for Data Agents

The

Tier 1 Calibration Run — World Model Capability Benchmark (Braitenberg Agent)

Tier 1 Calibration Run — World Model Capability Benchmark (Braitenberg Agent)

First calibration run of a cross-episode memory

The Art & Science of Benchmarking Agents — Vincent Chen, Snorkel AI

The Art & Science of Benchmarking Agents — Vincent Chen, Snorkel AI

ARC AGI 3 launched a few weeks before this talk with every task human solvable and frontier models under 1%. That gap is the ...

Agent Arena: The Benchmark That Shows How AI Agents Actually Fail

Agent Arena: The Benchmark That Shows How AI Agents Actually Fail

Agent

Agents' Last Exam: Benchmarking AI Agents on 1K+ Real-World Economic Tasks

Agents' Last Exam: Benchmarking AI Agents on 1K+ Real-World Economic Tasks

This video unpacks

NVIDIA Blackwell Leads AgentPerf: 20x Agentic AI Infrastructure Performance Explained

NVIDIA Blackwell Leads AgentPerf: 20x Agentic AI Infrastructure Performance Explained

NVIDIA Blackwell delivered leading results in the

We built our own agent benchmark (and you should too)

We built our own agent benchmark (and you should too)

Most

LocalSearchBench: Benchmark for Agentic Local Search

LocalSearchBench: Benchmark for Agentic Local Search

In this AI Research Roundup episode, Alex discusses the paper: 'LocalSearchBench:

Evaluating AI Agents: Why Memory-Wipe Benchmarks Hide Real Learning

Evaluating AI Agents: Why Memory-Wipe Benchmarks Hide Real Learning

Most AI

AI Perf benchmarking - Dynamo and other LLM endpoints

AI Perf benchmarking - Dynamo and other LLM endpoints

Join us as we cover features of Dynamo and walk you through a hands-on demo. See how Dynamo accelerates inference for ...