# Xplore AI launches engineering benchmark, with GPT-6 Astra at 37%

> Source: <https://runtimewire.com/article/xplore-ai-rwe-bench-engineering-models>
> Published: 2026-10-27 17:21:58+00:00

# Xplore AI launches engineering benchmark, with GPT-6 Astra at 37%

**The benchmark tests models on CAD design, circuit-board routing and turbine simulations, with every requirement checked by deterministic engineering verifiers.**

        By [Ryan Merket](https://runtimewire.com/author/ryan-merket)
        · Published 

Primary source: [x.com](https://x.com/XploreAi/status/2115108667418964246)

## Why it matters

Xplore is trying to turn engineering expertise into two products: a way to measure model performance and reinforcement-learning environments to improve it. The 37% launch score offers a baseline, while the business depends on whether expert-built tasks can become useful training infrastructure.

[Xplore AI](https://xplore.ai/) introduced its Real World Engineering Bench on October 27th, 2026, a benchmark that tests AI agents on constrained mechanical, electrical and aerospace design tasks. In its [launch thread on X](https://x.com/XploreAi/status/2115108667418964246), Xplore said it evaluated 11 models across 20 tasks and that [GPT-6 Astra](https://runtimewire.com/models/native-openai/gpt-6-astra-576a599a64cb3dde) had the highest average pass rate, at 37%.

When you purchase through links in our articles, [we may earn a small commission](https://runtimewire.com/affiliate-monetization-standards). This doesn’t affect our editorial independence.