- Published on
OpenAI cuts GPT-5.6 Luna prices by 80% as model competition shifts toward cost
- Authors
- Name
- DeepHorus
OpenAI has ignited a new phase in the AI price war by slashing the prices of two models in its GPT-5.6 frontier series, cutting GPT-5.6 Luna by 80% and GPT-5.6 Terra by 20%, while introducing a premium Fast mode for its flagship GPT-5.6 Sol. The aggressive pricing moves, announced by CEO Sam Altman on X as 'major price cuts today,' aim to undercut rivals such as Google and Anthropic as the competitive landscape shifts from raw capability to cost efficiency. These reductions come just days after Anthropic released Claude Opus 5 at the same price as Opus 4.8 and Google launched Gemini 3.6 Flash and Gemini 3.5 Flash-Lite, both built around lower inference costs.
The most dramatic change is Luna's price cut. Previously priced at $1 per million input tokens and $6 per million output tokens for a combined $7, Luna now costs $0.20 per million input tokens and $1.20 per million output tokens, totaling just $1.40 per million tokens. This places Luna below Google's Gemini 3.5 Flash-Lite ($2.80 combined) and far below Gemini 3.6 Flash ($9). While not the cheapest overallâXiaomi's MiMo-V2.5 Flash and DeepSeek's flash model remain cheaperâthe reduction brings an OpenAI frontier model into direct competition with the low-cost inference tier.
Terra's 20% cut moves its combined price from $17.50 to $14 per million tokens, matching Google's Gemini 3.1 Pro Preview pricing for contexts under 200,000 tokens. Terra also undercuts OpenAI's own GPT-5.4 ($17.50 combined), offering comparable intelligence for about 1/13th the cost. Meanwhile, Sol Standard pricing remains unchanged at $5 input and $30 output per million tokens, but OpenAI adds Sol Fast mode at double the price ($10 input, $60 output), promising up to 2.5 times throughput without altering the model's intelligence.
The price cuts signal a strategic shift toward affordability and speed as AI model competition intensifies. OpenAI is successfully undercutting Google's price per intelligence and attempting to sway Anthropic users with faster performance. The GPT-5.6 series, first rolled out in late June 2026 via a limited U.S. government request, targets different workloads: Sol for complex reasoning and agentic tasks, Terra for general production use, and Luna for high-throughput, low-latency applications like summarization and real-time assistants.
Industry observers note that these moves could pressure competitors to further lower prices, especially as enterprises increasingly prioritize cost over marginal intelligence gains. VentureBeat's Frontier Model API pricing comparison highlights that OpenAI's Luna now occupies a sweet spot between frontier capability and low cost. However, analysts caution that the price war may squeeze margins for smaller AI firms and accelerate consolidation.
Looking ahead, OpenAI's pricing strategy may set a precedent for the industry. With Luna now competitive against low-cost providers, and Terra matching mid-tier rivals, the battle is moving toward inference efficiency and application-specific optimizations. The addition of Sol Fast mode suggests OpenAI is also targeting latency-sensitive use cases. As more players like Xiaomi and DeepSeek enter the market, further price reductions and specialized model offerings seem inevitable, reshaping the AI ecosystem toward cost-driven innovation.
