Enterprise AI Costs Hit 2026 Low Amid Price Wars and Chinese Open-Source Models
Jefferies, an investment bank, has found that the costs of operating AI models for companies have hit a yearly low due to global price competition and affordable Chinese open-source tools like those from DeepSeek. Between August 6 and 8, average inference costs were between $1.16 and $1.18 per million tokens, marking the lowest price this year. This is a significant drop from $2.04 on May 31 and $1.45 in late July. Jefferies shared this information on Monday, citing Silicon Data, a firm that tracks pricing across various business API providers. Analysts, led by Thomas Chong, believe this trend is part of a broader push for cost efficiency in both the US and Chinese tech industries. Last month, OpenAI also cut prices for its latest GPT-5.6 models.
Key facts
- Average inference prices ranged between US$1.16 and US$1.18 from August 6 to 8.
- This is the lowest level recorded in 2026, according to Jefferies.
- Prices have dropped from US$2.04 on May 31 and US$1.45 in late July.
- Silicon Data's index tracks pricing across business API providers and open-weight inference platforms.
- Jefferies analysts led by Thomas Chong noted an 'increasing emphasis on cost efficiencies'.
- OpenAI slashed rates for its GPT-5.6 model series last month.
- Chinese firms are pushing affordable computing through open-source models like DeepSeek.
Entities
Institutions
- Jefferies
- Silicon Data
- OpenAI
- DeepSeek
Locations
- United States
- China