OpenAI cuts GPT-5.6 Luna API pricing by 80% to outpace Chinese AI rivals
Synopsis
OpenAI has cut GPT-5.6 Luna API pricing by 80% — to just US$0.20 per million input tokens — instantly vaulting it to the top of intelligence-per-dollar rankings above Chinese rivals Zhipu AI and MiniMax, in the starkest sign yet that the global AI pricing war has reached the frontier model tier.
Key Takeaways
OpenAI cut GPT-5.6 Luna API pricing by 80 per cent on 31 July 2026 , to US$0.20 per million input tokens and US$1.20 per million output tokens .
The mid-tier GPT-5.6 Terra model was reduced by 20 per cent , now priced at US$2 per million input tokens and US$12 per million output tokens .
The flagship GPT-5.6 Sol received a new 'fast mode' delivering up to 2.5x performance speed improvement.
Research firm Artificial Analysis ranked GPT-5.6 Luna as the top model in intelligence-per-dollar metrics, above Zhipu AI 's GLM-5.2 and MiniMax 's M3 .
The announcement was made by CEO Sam Altman on social media platform X and reflects mounting competitive pressure from lower-cost Chinese AI alternatives including DeepSeek and Kimi .
OpenAI has slashed developer pricing for its GPT-5.6 Luna model by 80 per cent, bringing API costs down to US$0.20 per million input tokens and US$1.20 per million output tokens, in a direct response to intensifying competition from lower-cost Chinese artificial intelligence alternatives. OpenAI CEO Sam Altman announced the cuts on Thursday, 31 July 2026, via a post on X, marking one of the company's most aggressive pricing moves to date.
What was announced
The price reduction covers three models from the GPT-5.6 family, all introduced earlier this month. Beyond the 80 per cent cut on GPT-5.6 Luna, the mid-tier GPT-5.6 Terra model saw a 20 per cent reduction, now priced at US$2 per million input tokens and US$12 per million output tokens. The flagship GPT-5.6 Sol received a new 'fast mode' that the San Francisco-based company said would boost performance speed by up to 2.5 times.
API fees represent what developers pay to embed AI capabilities directly into their own applications and services — making pricing a critical lever in the race to capture developer mindshare globally.
The competitive backdrop
The cuts come as open-weight and accessible Chinese AI models continue to win over international developers through aggressive pricing and rapidly improving capabilities. Research firm Artificial Analysis immediately ranked GPT-5.6 Luna as the most attractive model in its intelligence-per-dollar rankings following the announcement, placing it above competitors including Zhipu AI's GLM-5.2 and MiniMax's M3.
The move signals that even the most capitalised Western AI lab is not immune to margin pressure from Beijing-backed and independent Chinese challengers such as DeepSeek, Moonshot AI's Kimi, and others that have undercut incumbent pricing for months.
Why it matters
For developers and enterprises evaluating AI infrastructure costs, an 80 per cent price drop on a frontier-class lightweight model is a material shift. It lowers the barrier for startups and mid-market companies to deploy capable AI at scale without the cost burden previously associated with top-tier US models.
Rivals including Anthropic's Claude Fable and others in the space will now face renewed pressure to respond with their own pricing adjustments or capability differentiation to retain developer loyalty.
What's next
Industry analysts will be watching whether this round of cuts triggers a broader repricing across the AI API market, particularly from Anthropic and other Western incumbents. The sustainability of such deep discounts — and whether they compress OpenAI's margins ahead of a potential public listing — remains a key question for investors and observers tracking the company's financial trajectory.
Point of View
It implicitly validates the capability claims of rivals like DeepSeek and MiniMax that Western incumbents spent months dismissing. The intelligence-per-dollar framing adopted by Artificial Analysis is becoming the dominant competitive metric, shifting power toward developers and away from model providers. For Anthropic and others still pricing at premium tiers, the window to respond without a revenue hit is narrowing fast.
NationPress
31 Jul 2026
Frequently Asked Questions
How much did OpenAI cut GPT-5.6 Luna pricing?
OpenAI cut GPT-5.6 Luna API pricing by 80 per cent , bringing costs down to US$0.20 per million input tokens and US$1.20 per million output tokens . The announcement was made by CEO Sam Altman on 31 July 2026 .
Why did OpenAI slash its AI model prices?
OpenAI reduced prices to defend its market share against rapidly advancing, lower-cost Chinese AI rivals. Models from companies such as Zhipu AI , MiniMax , and DeepSeek have been winning over international developers with aggressive pricing and improving capabilities.
Which OpenAI models received price cuts in July 2026?
Three models from the GPT-5.6 family were updated: GPT-5.6 Luna saw an 80 per cent price cut, GPT-5.6 Terra was reduced by 20 per cent , and GPT-5.6 Sol received a new 'fast mode' boosting speed by up to 2.5 times .
How does GPT-5.6 Luna compare to Chinese AI models on price?
Following the price cut, research firm Artificial Analysis ranked GPT-5.6 Luna as the most attractive model in its intelligence-per-dollar rankings. It placed above Chinese rivals including Zhipu AI 's GLM-5.2 and MiniMax 's M3 .
Will Anthropic and other AI companies cut prices in response?
OpenAI 's aggressive repricing is expected to put pressure on rivals including Anthropic and its Claude Fable model line. Industry analysts are watching whether a broader repricing wave follows across the AI API market in the coming weeks.
