OpenAI released GPT-5.6 Sol to general users today, alongside two smaller models Terra and Luna, after the U.S. Department of Commerce restricted a preview to about 20 trusted partners for two weeks. Sol in ultra mode scored 91.9% on Terminal-Bench 2.1 and matched Anthropic's restricted Mythos Preview on ExploitBench while using roughly one-third of the tokens. The launch follows Grok 4.5 by one day and Meta's Muse Spark 1.1 by hours, intensifying competition among frontier AI models as Google's November 2025 Gemini 3 remains the oldest flagship still in active deployment.
OpenAI Introduces Three-Tier Pricing Structure for Sol, Terra, and Luna
OpenAI priced Sol at $5 and $30 per million input and output tokens, with Luna at $1 and $6 per million tokens. Terra matches GPT-5.5 performance at half the price, according to OpenAI. The company introduced two new configuration options: a max reasoning effort setting that extends Sol's processing time, and an ultra mode that distributes work to subagents. For context, Anthropic charges $10/$50 for Claude Fable 5, Google charges $2/$12 for Gemini 3.1 Pro, and xAI charges $15/$75 for Grok 4.5. DeepSeek charges $1.74/$3.48 for V4 Pro, and Xiaomi charges $1/$5 for MiMo v2.5 Pro.
Sol Achieves 91.9% on Terminal-Bench 2.1 and Matches Mythos Preview on ExploitBench
On Terminal-Bench 2.1, which measures command-line workflow completion, Sol in ultra configuration scored 91.9%, with standard Sol at 88.8%. These results place Sol ahead of Anthropic's Claude Mythos 5 at 88.0%, Claude Fable 5 at 84.3%, and Claude Opus 4.8 at 78.9%. Google's Gemini 3.1 Pro Preview scored 70.7%. On ExploitBench, which tests software vulnerability identification and exploitation, Sol matched the restricted Mythos Preview while spending roughly one-third of the tokens. OpenAI stated Sol does not cross the "Cyber Critical" line in its risk framework.
Early Testers Compare Sol Performance to Claude Fable 5
Theo, CEO of AI platform T3 Chat, called Sol "world leading in computer use" and said it fixed complaints he had with GPT-5.5. Dan Shipper, whose team at Every tested Sol for about a month, stated "GPT-5.6 is like a Porsche, Fable is like a warp drive," positioning Sol as a daily driver and Fable for more demanding tasks. Researcher Daichi Konno reported Sol clearly beats GPT-5.5 and lands near Fable 5, with Anthropic still ahead in writing tasks. Konno noted Sol's safeguards did not activate on life-science questions, which he believes could make it a default for biology work.
OpenAI Release Follows Grok 4.5 and Meta Muse Spark 1.1 Launches
Sol's release came one day after SpaceXAI shipped Grok 4.5 and hours after Meta released Muse Spark 1.1, its first paid model. Elon Musk called Grok 4.5 "roughly comparable to Opus 4.7, but much faster." Anthropic's Fable 5 moved to usage credits only once its 50% weekly allowance expired on July 7, after returning globally on July 1. Google's Gemini 3, released in November 2025, remains the oldest frontier flagship among major U.S. labs without a recent refresh.
FAQ
What did OpenAI release today?
OpenAI released GPT-5.6 Sol to general users today, alongside two smaller models Terra and Luna, after a two-week preview restricted by the U.S. Department of Commerce to about 20 trusted partners.
How does Sol perform on Terminal-Bench 2.1 compared to other models?
Sol in ultra mode scored 91.9% on Terminal-Bench 2.1, ahead of Claude Mythos 5 at 88.0%, Claude Fable 5 at 84.3%, Claude Opus 4.8 at 78.9%, and Google's Gemini 3.1 Pro Preview at 70.7%. Standard Sol scored 88.8%.
What is the pricing for OpenAI's new models?
Sol costs $5 and $30 per million input and output tokens, while Luna costs $1 and $6 per million tokens. Terra is priced to match GPT-5.5 performance at half the cost, according to OpenAI.
