Price Wars in AI: OpenAI Slashes GPT-5.6 Luna Costs by 80% Amidst Competitive Market Shift

In a notable shift within the AI landscape, OpenAI has announced significant price reductions for two models in its GPT-5.6 series. The GPT-5.6 Luna, recognized as the smallest and quickest in the lineup, sees an 80% decrease in price, while the mid-tier GPT-5.6 Terra will drop by 20%. The flagship GPT-5.6 Sol model will now include a premium Fast mode at a higher cost.

This pricing adjustment makes Luna more competitive against the most affordable commercial models on the market. The changes come just after Anthropic unveiled its new Claude Opus 5 at the same price point as the previous Opus 4.8, and Google launched its Gemini 3.6 Flash and Gemini 3.5 Flash-Lite models, which focus on cost efficiency and rapid execution.

OpenAI aims to undercut Google’s pricing strategy and entice users from Anthropic by offering superior speed with its latest models. The new pricing for Luna stands at $0.20 per million input tokens and $1.20 per million output tokens, totaling $1.40 per million tokens. The costs for Terra are set at $2 per million input tokens and $12 per million output tokens, totaling $14 per million.

The pricing for Sol Standard remains the same at $5 per million input tokens and $30 per million output tokens. The newly introduced Sol Fast mode will be priced at $10 per million input tokens and $60 per million output tokens, providing a throughput increase of up to 2.5 times without altering the base model’s intelligence.

Sam Altman, co-founder and CEO of OpenAI, shared these updates on social media, branding the adjustments as “major price cuts.”

Comparison of AI Model API Pricing

Pricing is quoted per million tokens, with total costs being input plus output prices. Cached-input costs are not included for a consistent comparison.

OpenAI Adjusts Luna Pricing to Compete

The most impactful change is the substantial price reduction for Luna. Originally priced at $1 per million input tokens and $6 for output tokens, the new combined rate has dropped to $1.40. This new rate places Luna below Google’s Gemini 3.5 Flash-Lite, priced at $2.80 for both input and output combined, and significantly beneath Gemini 3.6 Flash at $9.

While Luna is not the cheapest model in the market overall, it is positioned to compete more directly with entry-level models such as Xiaomi’s MiMo-V2.5 Flash, which has lower costs on a per token basis. The reduction signals OpenAI’s intent to penetrate the low-cost inference model segment more aggressively.

The GPT-5.6 series was launched in June 2026, with models designed to address varying needs for intelligence, speed, and cost. As such, Sol caters to highly complex tasks, Terra balances efficiency and capability for general applications, while Luna is optimized for high-speed, low-cost operations.

Terra’s Pricing Aligns with Competitors

With a 20% price decrease, Terra now has a total price of $14 per million tokens, bringing it in line with Google’s Gemini 3.1 Pro Preview for similar token lengths. The lower price also positions Terra under OpenAI’s own GPT-5.4 pricing, offering customers comparable intelligence at a much lower cost.

This adjustment further differentiates the three tiers within OpenAI’s GPT-5.6 models. For instance, Luna is priced at one-tenth of Terra’s cost per request, and Terra remains significantly more affordable than Sol Standard.

On a contrasting note, the Sol Fast mode is priced at a premium of $70 per million tokens, establishing it as the most expensive option in the discussed models, reflecting OpenAI’s strategy to maintain higher pricing for latency-sensitive tasks.

New Prices Emerge Amid Competitive Landscape

These pricing changes were announced shortly after Google launched its own budget-friendly offerings with Gemini 3.6 Flash and Gemini 3.5 Flash-Lite, which emphasize minimizing token usage and enhancing execution speed.

Google’s new models are positioned as economical choices for long-term software development and knowledge work, emphasizing efficiency in token consumption and performance.

Despite these competitive adjustments, OpenAI’s models have reportedly outperformed Google’s in independent analyses, particularly with Luna exceeding the capabilities of both Gemini 3.6 Flash and 3.1 Pro models, making OpenAI’s offerings potentially more favorable in terms of cost-efficiency.

The rapid changes in pricing strategies highlight the competitive nature of the AI model market. OpenAI’s latest revisions represent a significant shift in model pricing rather than a new generation of models.

Shifting Focus in AI Market Dynamics

These price reductions signal a transition in the competitive landscape, indicating that access to advanced capabilities is no longer the sole differentiator. Businesses must now consider operational costs linked to deploying these models effectively.

While OpenAI may not hold the title of the lowest-cost option, the substantial cut in Luna’s pricing has positioned it to compete more directly with smaller providers such as Google and Xiaomi. This is especially crucial for applications that require high-volume token usage, where small pricing differences can significantly impact overall expenses in various operational contexts.

Ultimately, OpenAI’s positioning of the GPT-5.6 series suggests a strategic realignment with Sol as the premium choice, Terra as a contender in the professional tier, and Luna as a direct competitor to budget-friendly models in the industry.