AI Model Report

Reviews · AUGUST 23, 2026

OpenAI cuts GPT-5.6 Sol API pricing more than 20% for three months

Sol's input drops to $4 per million tokens and output to $20 for a 90-day promotional window, undercutting Claude Opus 5 on output and pulling the frontier price floor down for every tool built on the API.

By Karl Strauchman · Senior model reviewer · August 23, 2026

OpenAI cut developer pricing on GPT-5.6 Sol by more than 20% on August 21, 2026, according to a Reuters report carried by Investing.com. Input tokens drop from $5 to $4 per million (a 20% cut) and output tokens from $30 to $20 per million (a 33% cut), for a 90-day promotional window. The lower rate holds through late November.

The comparison that matters: Claude Opus 5 sits at $5 input and $25 output per million tokens. Sol is now cheaper than Anthropic's flagship on both sides of the ledger, and materially cheaper than Claude Fable 5, which lists at $10 input and $50 output. Sol also posts a 53.6 on Agents' Last Exam against Fable 5's 40.5, and an 80 on the Artificial Analysis Coding Agent Index.

This is the second frontier repricing in three weeks. On July 30, OpenAI cut GPT-5.6 Terra by 20% and GPT-5.6 Luna by 80%, a move we covered when it landed. The Sol cut extends the same posture up the intelligence curve. Credit allocations for ChatGPT Work and Codex adjust accordingly.

Downstream, the discount compounds. Vercel's AI Gateway had already listed Sol at $2.50 input and $15 output per million, below OpenAI's own list; it's now moved to the new $4/$20 base and layered a 50% promotion on top, bringing gateway customers to $2 input and $10 output through September 18. That's a working ceiling on what any competitive router can charge for equivalent traffic.

The commercial backdrop explains the aggression. Ramp's spend data across more than 70,000 U.S. businesses showed Anthropic at roughly 44% share against OpenAI's near-40% as of July, with OpenAI growing faster into Q3. Ramp economist Ara Kharazian described Sol as "increasingly the choice for developers." Price is the lever OpenAI can pull without waiting on a model release, and it's pulling it repeatedly.

For the five-to-thirty-person companies running proactive outreach, content, and lead-qualification on top of frontier APIs, the arithmetic changes today. A 33% cut on output tokens is a 33% cut on the expensive half of most generation workloads. Tools that were marginal at $30 output are comfortable at $20. Services like LemonLime, which prepares finished sales and marketing work on frontier infrastructure and delivers it proactively rather than making the owner operate another dashboard, sit directly downstream of this cost floor. When the floor drops for 90 days, the practical cost of done-for-you AI customer acquisition drops with it.

The window is dated. Whether the promotional rate becomes the new list price or reverts on November 21 is the question every buyer routing production traffic should be pricing in now. OpenAI hasn't said. It rarely does until the next cut.

Sources

  • https://openai.com/index/gpt-5-6/
  • https://www.investing.com/news/stock-market-news/openai-cuts-developer-pricing-for-frontier-gpt56-sol-model-by-more-than-20-4872186
  • https://vercel.com/changelog/gpt-5-6-sol-is-now-50-percent-off-a-lower-price
  • https://techcrunch.com/2026/08/20/openai-is-gaining-on-anthropic-with-business-users-new-data-indicates/
  • https://www.cnbc.com/2026/07/30/open-ai-price-cut-gpt.html