Why would smaller and worse models not be 80% cheaper?
If I can run those models on my consumer hardware, I'd better believe they are 80% cheaper than the models that need 1 TB of RAM.
Comment from poster says they are offering Deepseek v4-Pro. Cannot find any details on website.
For the very hardest reasoning tasks GPT-5 and Opus are still ahead, no argument there. But what we see in practice is customers dropping in an open-source model and getting very similar results on 80-90% of real-world use cases — with significant cost savings and end-to-end UK data residency (which matters a lot for our enterprise and institutional customers). And on the consumer-hardware point: these are Blackwell GPUs in a UK datacentre, in a token factory architecture.