Good news for those like me who use GPT! OpenAI has significantly reduced API prices starting today, mainly due to server-side optimizations that freed up cost space. This time, improvements to the production line GPU kernel have lowered service costs by 20%, and speculative decoding has boosted token generation efficiency by over 15%. Key highlights: - GPT-5.6 Luna: Prices slashed by 80%, now $0.20 per million input tokens and $1.20 for output tokens. - Terra: Price cut by 20%, adjusted to $2 for input and $12 for output. - Sol: Standard pricing remains $5 / $30, but a new "High-Speed Mode" has been added, with speeds up to 2.5x faster at double the standard price. - Auto Review (the "Auto Review" feature in apps): Now powered by Luna, expected costs are reduced to one-tenth of the original. Additionally, this price drop will directly reflect in usage calculations for Codex and ChatGPT Work (enterprise edition), meaning you can now run longer with the same quota.