OpenAI cuts GPT-5.6 Sol pricing by over 20%

OpenAI is reducing the API and credit pricing of its flagship GPT-5.6 Sol model by more than 20%, according to a Reuters report on Friday. The move marks the first price cut for the company's most capable reasoning model since the GPT-5.6 family launched in July.tradingview

Sol Joins the Price Cut Trend

Until now, Sol had been the lone holdout in OpenAI's GPT-5.6 pricing lineup. When the company slashed prices on July 30, it cut GPT-5.6 Luna by 80% and GPT-5.6 Terra by 20%, while explicitly leaving Sol's rates unchanged at $5 per million input tokens and $30 per million output tokens. Sol is positioned as the top-tier model for complex reasoning, coding, and agentic workflows.openai+1

The new reduction brings Sol's pricing more in line with competitive pressure from third-party providers. OpenRouter, for example, had already been offering Sol at $2.50 per million input tokens — a 50% discount to OpenAI's direct price — by routing requests across multiple infrastructure providers including Amazon Web Services Amazon.com, Inc. ↘ 0.57% and Microsoft ↗ 0.43% Azure.openrouter

Amazon Bedrock Expands GPT-5.6 Access

In a related development, AWS on Wednesday introduced cross-region inference support for all three GPT-5.6 variants — Sol, Terra, and Luna — on Amazon Bedrock. The feature allows requests to route across more than 25 AWS regions globally, drawing on a broader pool of compute capacity to improve throughput under heavy load.aws.amazon

The Bedrock integration supports both geographic inference profiles, which keep data processing within a single geography, and global profiles that route to any supported commercial region. All three models accept text and image inputs, offer a one-million-token context window, and can be called through OpenAI's native Responses and Chat Completions APIs.aws.amazon

Intensifying Price Competition

The Sol price cut continues a pattern of rapid cost declines across frontier AI models in 2026. OpenAI's July 30 reductions were framed as the result of the company using GPT-5.6 itself to optimize runtime efficiency, yielding 20% lower serving costs through production GPU kernel improvements. The broader availability through Amazon Bedrock adds further competitive pressure by giving enterprise customers alternative procurement channels with built-in failover and regional compliance options.openai