AI Price War: OpenAI Cuts GPT-5.6 Sol Output Pricing for Developers by One-Third
·
22.08.2026 10:40
·
2 min read
OpenAI has announced a temporary price reduction for its flagship language model, GPT-5.6 Sol. As reported by Reuters, the discount is valid through November 21, 2026, and is aimed at developers of enterprise software and AI agents.
For standard requests with context up to 272,000 tokens, output text generation pricing dropped by 33.3%—from $30 to $20 per million tokens—while the input processing price fell by 20%—from $5 to $4 per million tokens.
New Pricing and Long Context
According to the official OpenAI API price list and developer documentation, rates for the long context window have also been significantly revised:
- Context over 272,000 tokens: input processing now costs $8 per 1 million tokens (down from $10), and response generation costs $30 per 1 million tokens (down from $45).
- Where discounts apply: the updated rates apply to access via the OpenAI API, the partner platform Amazon Bedrock, as well as credit usage in Codex and ChatGPT Work.
- User subscriptions: pricing for standard ChatGPT Plus, Pro, and Business plans, as well as their included message limits, remain unchanged.
A Blow to Competitor Anthropic
This marks the third price reduction for the GPT-5.6 lineup in the past month: OpenAI previously lowered prices for its mid-tier Terra model by 20% and slashed costs for the ultra-fast Luna model by 80%.
Discounting the flagship Sol model ramps up pressure on main competitor Anthropic. During the three-month promotion, GPT-5.6 Sol pricing ($4 / $20 per million tokens) is noticeably lower than competitors’ flagship models—such as Claude Opus 5 ($5 / $25) and Claude Fable 5 ($10 / $50)—encouraging startups and enterprise companies to move their agent infrastructure to the OpenAI stack.
