The next generation in intelligence for work

“GPT‑6 delivered some of the best and highest-signal recommendations we've seen. It effectively navigated through our entire codebase of over 80 million lines of code, wrote and analysed over one thousand queries against our data warehouse, and used over 21 different internal knowledge sources and community feedback.”

“Astra does more with dramatically less, and without sacrificing trust. Across five clean artifact cases, it used roughly half the output tokens of Opus 4.6 and 60% fewer than Opus 5, while staying grounded enough to separate evidence from speculation and admit when a tool failed. That combination of efficiency, precision, and honesty is the step change.”

“We put GPT‑6 on real work across Epic and it managed C++ at the scale of Unreal Engine, worked out complex problems faster than anything we've tested, and showed real nuance to support non-engineering teams work multimodally. We're seeing early signs of a model that unlocks more creative exploration with less handholding.”

“GPT‑6 Astra is an impressively capable and efficient model that shines within Perplexity's codegen-native environments. Combining Astra with Perplexity's Search as Code architecture yields 9% higher performance versus prior models on our most difficult research benchmark, at just 49% of the cost.”