OpenAI's latest GPT-5.5 model introduces improved token efficiency, delivering enhanced output with fewer tokens, but users face steep price increases per million tokens compared to the prior generation.
- GPT-5.5 token prices up to two times higher than GPT-5.4
- Efficiency gains reduce tokens but not overall costs
- Anthropic’s Claude Opus 4.7 shows similar cost dynamics
What happened
OpenAI recently launched its GPT-5.5 model, which features improved intelligence and more efficient processing of tokens. The company raised its pricing significantly, with input tokens costing $5 per million, cached input at $0.50, and output tokens at $30 per million. These rates represent roughly double those of the prior GPT-5.4 model.
Despite the higher pricing, GPT-5.5 uses fewer tokens, especially on longer prompts exceeding 10,000 tokens, generating between 19% and 34% fewer completion tokens. However, for shorter prompts, the reduction in token usage is less pronounced, resulting in steeper cost increases for many users.
Why it matters
This pricing shift highlights the escalating cost pressure faced by AI developers, partly driven by growing computational demands and investment losses. OpenAI is reportedly anticipating a $14 billion loss in 2026, necessitating increased revenue from usage fees. This environment encourages customers to weigh efficiency gains against expanded expenditure.
Other major AI providers face similar pressures. Anthropic’s recent Claude Opus 4.7 model, which claimed improved tokenization, did not announce a price increase but analysis reveals increased costs for longer prompts. These trends suggest that rising prices may become the norm as frontier AI models grow more capable yet resource-intensive.
What to watch next
Market watchers should track how customers adapt to these rising expenses, particularly enterprise users relying on large-scale generative AI workflows. Developments in tokenizer technology and prompt engineering could help mitigate cost increases by improving token efficiency further.
Additionally, OpenAI and competitors’ strategies for balancing model improvements with pricing will be critical. Pricing adjustments could influence adoption rates and encourage alternative AI platforms or open-source options, especially if operational losses continue at high levels.