AI Agents for Business: Test Results
Artificial Analysis has introduced AA-AgentPerf, the industry's first open benchmark that measures inference system performance on real agentic coding tasks. Unlike conventional tests, it profiles an agent's full multi-step trajectories: writing code, running it, analyzing errors and iterating. NVIDIA took the leading position on the key metrics of the new standard.
AI-processed from NVIDIA Developer Blog; edited by Hamidun News
The first open benchmark of AI agents for business, AA-AgentPerf, has been released. For the first time, it gives an objective measure of system performance on real automation tasks, and NVIDIA took the lead in this multi-vendor standard.
Which AI agent system is the fastest for automation?
According to AA-AgentPerf, the first open benchmark of its kind, NVIDIA has for the first time taken the lead in inference system performance on real agentic coding tasks.
Why does AI agent testing need a dedicated benchmark?
AI agents load infrastructure differently from ordinary requests. Older benchmarks (tokens per second) do not reflect real-world performance, so the industry needs a tool like AA-AgentPerf.
- Nov 1, 2026Local models lose to Opus and Codex: 0.21 on a MacBook versus 1.00 for Codex
- Nov 1, 2026Adobe's EditAgent judge picks a usable edit 81.3% of the time, versus 74.8% for EditScore
- Nov 1, 202616 models tested on 7 search tools: Exa averaged 10.5 out of 14, SearXNG about 3
- Nov 1, 2026Google's StreamVibe video agent beats other models by 10.7% on response timing and quality
Key Questions Answered
Which AI agent system is the fastest for automation?
According to AA-AgentPerf, the first open benchmark of its kind, NVIDIA has for the first time taken the lead in inference system performance on real agentic coding tasks.
Why does AI agent testing need a dedicated benchmark?
AI agents load infrastructure differently from ordinary requests. Older benchmarks (tokens per second) do not reflect real-world performance, so the industry needs a tool like AA-AgentPerf.
Test Frontier LLMs & Autonomous Agents in One Workspace
Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.