Opus 5.5 and GPT-6 Sol: The model war moves to pricing

New models from Anthropic and OpenAI do not bring a capability breakthrough. They bring price cuts. That is now the main front of competition. Opus 5.5, the latest version of Anthropic's main model for coding and complex knowledge work, costs 4 dollars per million input tokens and 20 dollars per million output tokens. That is 20 percent less than Opus 5. Cache reads, which make up most of the bill in agent work and coding, now cost 0.20 dollars per million tokens, down 60 percent. The model also answers more than 30 percent faster.

Anthropic says savings reach 40 percent on typical tasks with default settings because the model uses fewer tokens to finish the same job. One caveat applies to high-risk areas like cybersecurity and biology. Queries from those fields may be routed automatically to the older model.

OpenAI responds with GPT-6 Sol and Luna, lighter variants in the GPT-6 family. Sol is an everyday tool for tasks needing decent quality at a lower cost. Luna is cheaper and smaller. The companies are racing not only against each other but against open models. Businesses increasingly use model routers to rely on expensive frontier models less often.

In China, the same mechanism appears in DeepSeek's price list. From 10 September, 12:00 Beijing time, the flash model runs at new rates. Input with a cache hit fell in off-peak hours from 0.05 to 0.02 yuan per million tokens, a 60 percent drop. Input without a hit fell from 1.5 to 1 yuan, one third less. Output fell from 4.5 to 4 yuan, about 11 percent less. Peak-hour rates are twice the off-peak ones.

The cut is notable because on 17 August DeepSeek introduced peak tariffs and raised the output price from 2 yuan to 4.5 yuan in off-peak hours and 9 yuan at peak. That was a 125 percent and 350 percent increase. A little over a month later, the company realized it could cut prices without losing revenue. On 19 September, it clarified that weekends and public holidays in China are billed around the clock at off-peak rates, lowering bills for many teams.

For customers, the direction of rate changes matters more than the level. A price cut on top-tier models changes the math not only in labs but for everyone building products on them. What was an experiment a year ago now fits a small team's budget. The price war does not mean models get cheaper forever. Suppliers shift costs to weaker versions and peak-hour terms. Anyone who moves load into the off-peak window pays less today than anyone paid a year ago for a comparable volume.