Revolut says that in less than a year it’s managed to slash its [code generation] costs by roughly 90 percent by making self-hosted models its default instead of relying primarily on commercial services like Claude.
Revolut says that in less than a year it’s managed to slash its [code generation] costs by roughly 90 percent by making self-hosted models its default instead of relying primarily on commercial services like Claude.
About time this hit enterprise.
Been saying the downfall of OpenAI and Anthropic will be local models that can accomplish the same thing without needing a dstacenter worth of GPUs to run.
Plus you have no total token limit on self hosted models. You can let a slop agent run for weeks without ballooning costs or getting rate limited.
And no model censoring, so you can go build that nuclear bomb you always wanted lol.
Both AI giants are already losing a ton of money because their pricing isn’t viable for the amount of training they’re doing trying to be on top. If they readjust, they’re gonna suffocate against much cheaper solutions.
The data center is still needed, just not the same scale. You don’t run “small” local models to compete with Claude’s size. They don’t publish numbers, but estimates were like 1-2T params for the models like early Opus. You need a lot of hardware to run those efficiently at a scale an enterprise would need.
Or they just spent enough to upgrade developer desktops with RTX Pro’s to run quantized ~100B param models at their desks. Those are very good for coding.
Microsoft predicted this some time ago which is why they have begun pushing a Hybrid Model where tasks are run locally first and only going to the cloud for heavy lift.