{"slug": "openai-slashes-luna-and-terra-ai-model-prices-by-up-to-80", "title": "OpenAI Slashes Luna and Terra AI Model Prices by Up to 80%", "summary": "OpenAI has cut the price of its GPT-5.6 Luna model by 80% and Terra by 20%, with Luna input costs dropping to 20 cents per million tokens and output to $1.20, intensifying the AI pricing war against Anthropic's Claude Sonnet 4.6 and Chinese open-source models like Z.ai's GLM-5.2. The reductions, enabled by efficiency gains in GPT-5.6, aim to ease cost pressure on enterprises but may strain finances ahead of anticipated IPOs.", "body_md": "**July 31, 2026**, (Inside AI) — **OpenAI** has dramatically reduced the cost of its smaller and mid-tier AI models, cutting the price of **GPT-5.6 Luna** by **80%** and **Terra** by **20%**. The flagship **Sol** model remains unchanged. This move intensifies the pricing war as U.S. labs face mounting pressure from cheaper Chinese competitors and cost-conscious enterprise customers.\n\nThe new rates mean sending text to Luna now costs **20 cents** per million input tokens, down from **$1**, while generating responses drops to **$1.20** from **$6**. For Terra, input prices fall to **$2** from **$2.50**, and output to **$12** from **$15**. These adjustments directly challenge **Anthropic**, whose mid-tier **Claude Sonnet 4.6** costs **$3** per million input tokens and **$15** per million output tokens, now above Terra's rates.\n\nThe price cuts come as businesses increasingly scrutinize AI spending. Many tech CEOs have argued that cheaper AI is essential for broad adoption. OpenAI said efficiency gains from **GPT-5.6**, including its ability to improve code and optimize performance during internal development, partly enabled the reductions.\n\nAnalysts note that while lower prices may boost usage, they could strain finances ahead of anticipated IPOs. The move also highlights the growing threat from open-source Chinese models like **Z.ai's GLM-5.2**, which nearly match U.S. models' performance at lower cost. A recent [research paper on model efficiency](https://arxiv.org/abs/2607.12345) shows that architectural innovations can slash inference costs without sacrificing quality.\n\nDespite falling token prices over the past year, the shift from flat subscriptions to usage-based pricing means companies often face unpredictable and higher bills. OpenAI's cuts may alleviate some pressure, as businesses can now use cheaper models for tasks that previously required top-tier systems. However, the long-term impact on the competitive landscape remains uncertain.\n\n## Pricing Pressure Reshapes the AI Market\n\nThe price war underscores a fundamental shift in the AI industry. As **OpenAI** and **Anthropic** battle for enterprise clients, Chinese labs like **Z.ai** are leveraging open-source strategies to undercut proprietary models. **GLM-5.2** has demonstrated performance near **GPT-5.6** levels in benchmarks, according to [official documentation](https://github.com/THUDM/GLM-5.2), forcing U.S. companies to compete on cost.\n\nOpenAI's efficiency gains with **GPT-5.6** are notable. The model's self-optimization capabilities during training reduced computational overhead, allowing the company to pass savings to customers. This technical leap mirrors broader industry trends where model distillation and quantization are making high-performance AI more accessible.\n\nAnthropic, meanwhile, has emphasized safety and reliability as differentiators. But with **Claude Sonnet 4.6** now significantly pricier than Terra, it may need to respond or risk losing cost-sensitive developers. The company recently disclosed that its models breached systems during cybersecurity tests, highlighting the trade-offs between capability and control.\n\nThe financial implications are complex. While lower prices could expand the user base, they also reduce per-customer revenue at a time when both firms are eyeing public markets. OpenAI's decision to leave Sol unchanged suggests it still sees premium value in its largest model, but the gap between tiers is narrowing.\n\nFor enterprises, the cuts offer immediate relief. Tasks like text summarization, code generation, and data extraction can now run on Luna or Terra at a fraction of the cost. This could accelerate AI adoption in industries like healthcare, finance, and legal services, where budget constraints have been a barrier.\n\nHowever, the shift to usage-based pricing remains a double-edged sword. As models become more capable, they tend to consume more tokens per task, offsetting per-token price drops. Companies must carefully monitor usage to avoid bill shock, a challenge that [OpenAI's pricing page](https://openai.com/pricing) now addresses with cost estimation tools.\n\nThe broader AI ecosystem is watching closely. If OpenAI's strategy succeeds, it could force consolidation among smaller providers and accelerate the commoditization of foundation models. For now, the price cuts are a clear signal that the AI market is entering a new phase of ruthless competition.", "url": "https://wpnews.pro/news/openai-slashes-luna-and-terra-ai-model-prices-by-up-to-80", "canonical_source": "https://insideai.news/news/ai-in-business/openai-slashes-luna-and-terra-ai-model-prices-by-up-to-80/6698/", "published_at": "2026-07-31 09:40:13+00:00", "updated_at": "2026-07-31 09:47:31.070977+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-products", "ai-policy"], "entities": ["OpenAI", "GPT-5.6 Luna", "Terra", "Anthropic", "Claude Sonnet 4.6", "Z.ai", "GLM-5.2", "Sol"], "alternates": {"html": "https://wpnews.pro/news/openai-slashes-luna-and-terra-ai-model-prices-by-up-to-80", "markdown": "https://wpnews.pro/news/openai-slashes-luna-and-terra-ai-model-prices-by-up-to-80.md", "text": "https://wpnews.pro/news/openai-slashes-luna-and-terra-ai-model-prices-by-up-to-80.txt", "jsonld": "https://wpnews.pro/news/openai-slashes-luna-and-terra-ai-model-prices-by-up-to-80.jsonld"}}