GPT-6 Sol and Luna Cut OpenAI API Prices in Half
2 min readOpenAI released two new frontier models on Tuesday, and the headline is not a benchmark score. It is the invoice. GPT-6 Sol and GPT-6 Luna both landed with API pricing cut roughly in half against the previous GPT-5.6 tier, and OpenAI says the cut is permanent rather than a promotional rate.
What Came Before
For most of the past year, developers building on OpenAI faced a familiar trade. The strongest model was the expensive one, and the cheap model was the one you used when accuracy did not matter much. GPT-5.6 sat at the top of that ladder with promotional pricing that many teams assumed would eventually rise.
That assumption shaped budgets. Companies capped how many requests they routed to the best model and fell back to weaker tiers for high volume work such as classification, summarization, and routine customer replies.
What OpenAI Announced
GPT-6 Sol runs at $2 per million input tokens and $10 per million output tokens. GPT-6 Luna, the lighter sibling, runs at $0.10 and $0.50. According to TechCrunch, an OpenAI spokesperson confirmed the pricing is not promotional.
On capability, OpenAI says Sol makes about half as many mistakes as its predecessor, reaching the reliability of the company’s Astra line at a far lower cost. The company credits training methods borrowed from Astra for gains in professional work, factuality, coding, computer use, and alignment. Luna is tuned for speed and volume instead of deep reasoning.
Both models are live for Plus, Pro, Business, Enterprise, and Edu users through ChatGPT Work and Codex, with API access alongside. Free and Go users can try Luna on desktop.
Why It Matters
Price is the quiet lever that decides which AI products get built. When the strong model costs what the weak model used to, the fallback tier loses its reason to exist, and workloads that were priced out become viable overnight. Expect a wave of applications that simply route everything to the better model.
It also sharpens the competitive squeeze. Anthropic shipped Claude Opus 5.5 the same day and xAI released Grok 4.7 a day earlier, so the frontier labs are now trading releases within hours of each other rather than months.
Watch for how rivals respond on price rather than benchmarks. That is where the next move in this race is likely to land.