GPT-6 Sol vs Opus 5.5: per-million-token pricing compared
Sol's $2/$10 list price is half of Opus 5.5's, but coding workloads are mostly cache reads, where both cost $0.20 per million tokens, so the real gap is mainly on output.
if you're trying to pick between Opus 5.5 and GPT-6, here's what they cost per 1M tokens: > Opus 5.5 is $4 in / $20 out > GPT-6 Sol is $2 / $10 > GPT-6 Astra is $10 / $50 so on paper Sol is half the price of Opus and Astra is 2.5x but if you're using them for coding, most of what you're paying forl is actually cache reads since the agent re-reads your codebase on every step. and there Opus 5.5 and Sol are both $0.20 per 1M, while Astra is $1 so the real difference between Opus and Sol is mostly on output, not the 2x it looks like if you're on a Claude plan, just use Opus 5.5, it's already the default in Claude Code and you're not paying per token anyway. if you're paying API bills at scale, Sol is worth testing. and I'd only try Astra on really complex tasks