OpenAI has launched GPT-5.6 Sol, a flagship model that scores 59 points on the Intelligence Index, placing it just one point behind Anthropic‘s Claude Fable 5. Independent evaluation firm Artificial Analysis reports that Sol dominates the Coding Agent Index with 80 points and leads the AA-Briefcase benchmark for presentation tasks. Despite this performance, Sol costs $1.04 per task, which is roughly one-third of the price charged by Fable 5. OpenAI also introduced a cache-write fee structure and claims the model burns up to 54 percent fewer output tokens during agentic coding work. The company states that smaller variants, Terra and Luna, are priced at 50 and 80 percent less respectively, while input and output rates range from $1 to $30 per million tokens depending on the tier.
OpenAI is attempting to gain market share by forcing Anthropic into a pricing competition while competitors like Meta and xAI push rates lower with their own models. This strategy creates pressure on established labs to cut costs but risks a race to the bottom that could damage the industry. The new pricing tiers and efficiency gains offer a different angle of competition beyond raw intelligence scores.
* Sol costs $1.04 per task compared to Fable 5’s higher rate.
* Cache reads receive a 90 percent discount on the new fee structure.
* Output token usage drops by up to 54 percent for coding tasks.
