# Perplexity adds GLM 5.3 to Perplexity Computer

The model is built for long-context, multimodal agent workloads and outperforms GLM 5.2 on the WANDR research benchmark.

- Perplexity uses WANDR as its benchmark for large-scale, evidence-backed research.

## Why it matters

Perplexity Computer users can now select GLM 5.3 for long-context and multimodal research agent tasks.

## Sources

- [Perplexity: Perplexity adds GLM 5.3 to Perplexity Computer](https://x.com/perplexity_ai/status/2093410587338575936)

---

Summarized by dstilled on 2026-08-28. https://dstilled.ai/story/8d396a9a-6171-4310-9e33-d65fd92d6a53