Perplexity Computer has integrated OpenAI’s GPT-5.6 model family across its platform, making Terra the default engine for subagents and Luna the go-to for automations. The integration was completed by July 12, three days after OpenAI made GPT-5.6 generally available on July 9. GPT-5.6 is a three-tier family: Sol, priced at $5 per million input tokens and $30 per million output tokens, handles complex reasoning and coding; Terra, at $2.50 input and $15 output, serves as the middle option; Luna, at $1 input and $6 output, is optimized for speed. Perplexity selected Terra for subagent workflows involving research, coding, and project management, while Luna handles repetitive automation tasks. The article notes that Perplexity integrated GPT-5.6 in three days, whereas decentralized alternatives face latency from consensus mechanisms, token-gated access, and smart contract overhead. The tiered pricing creates a benchmark for crypto AI projects, as Luna’s $1 per million input tokens through a centralized API raises questions about decentralized networks’ cost structures, which often include gas fees and token staking. Perplexity routes workflow parts to different models based on complexity and cost, with Sol presumably available for deep reasoning. Crypto AI projects may find advantage in the economic layer, as Luna’s revenue flows to OpenAI, while a decentralized equivalent could flow to GPU providers with token holders governing pricing.