GPT-6 Prompt Caching Cuts Costs by Up to 90%
OpenAI has upgraded prompt caching for GPT-6 Sol and Luna, cutting cached input-token costs by up to 90% and reducing latency for developers building AI agents and long-context applications. The GPT-6 prompt caching system now offers higher default cache-hit rates, performance diagnostics and explicit breakpoints for controlling which prompt sections are stored. Developers can also change reasoning levels and available tools without invalidating cached context. OpenAI said GPT-6 API pricing is roughly 50% below GPT-5 promotional rates, potentially lowering costs for sustained conversations and multi-step agent workflows. The GPT-6 prompt caching upgrades are being extended across the GPT-6 API, ChatGPT Work and Codex. GitHub Copilot previously reported reducing fresh prompt processing by more than 50% across billions of requests using earlier OpenAI caching technology. Separately, Perplexity selected GPT-6 Sol as the default model for its Light preset in Effort Mode, which is currently available on the web. The developments may increase demand for cheaper AI infrastructure and intensify competition among model providers, but they have no direct cryptocurrency catalyst.
Neutral
The expected cryptocurrency-market impact is neutral because the article concerns OpenAI model efficiency, API pricing and Perplexity’s product integration, rather than a blockchain network, token, exchange or regulatory decision. In the short term, traders may react to the broader AI narrative, particularly if lower inference costs improve sentiment toward AI-related equities or crypto projects linked to decentralised computing. However, there is no direct change to crypto liquidity, token supply, network activity or institutional flows. The cost reductions could support long-term growth in AI-agent usage and demand for computing infrastructure, potentially benefiting some AI and decentralised-AI projects indirectly. Similar announcements about cheaper or more capable AI models have often produced short-lived sector rotations rather than sustained moves across the wider crypto market. Traders should therefore monitor AI-related token volume, funding rates, correlations with technology stocks and follow-up adoption data. Without evidence of on-chain usage, partnerships or capital commitments, the announcement is unlikely to create a durable bullish or bearish signal for major cryptocurrencies.