OpenAI cancelled the GPT-6.1 Sol model on September 28 following internal testing failures, leaving GPT-6 Astra as the company's primary offering. Pricing structures vary significantly across the current frontier models, with Astra and Claude Fable 5.1 priced at $10 per million input tokens and $50 per million output tokens, while Sol and Gemini 4 Argon cost one-fifth of that amount.
Operational costs for agents depend heavily on caching mechanisms. Astra charges $1.00 for cache reads, which is four times higher than Fable 5.1 and ten times higher than Sol. Additionally, Gemini 4 Argon features a 1 million token output limit, whereas the other models are restricted to 128 000 tokens per response.
Performance metrics show mixed results across benchmarks. Claude Fable 5.1 leads on the Artificial Analysis Intelligence Index with a score of 70 in coding-agent tasks, compared to 67 for Astra. While Fable 5.1 shows higher token consumption per task, its caching efficiency can lower costs for long-running agent loops compared to Astra.
Access to these models remains restricted for specific use cases. Gemini 4 Argon is currently limited to Fairwind cyber defenders. OpenAI restricts advanced offensive security capabilities in Astra to its Daybreak program, and Anthropic gates its unrestricted Claude Mythos 5.1 model behind trusted access programs.
