If I were buying one today, I'd most certainly go for the 512, but that goes for reasons beyond just the performance itself (256GB is just a bit tight these days, but that of course depends on the individual use case. I use XCode and Logic Pro X, so I might be a little biased here).
But as far as the performance differences, I question whether it's really that noticeable in real world use cases, and I think it's worth discussing in-depth to see. Benchmarking the SSD of course shows about a 2x speed increase on the 512GB configurations, but we know that this doesn't necessarily translate to a 2x real-world speed increase in all everyday applications. The CPU operates out of RAM and not directly off of the drive, so once the OS has loaded the data that is required into RAM, there shouldn't be a terribly big performance impact. The SSD speeds do, of course, benefit application startup and bootup times, but even here, much of this might not necessarily storage bound at the kinds of speeds we are used to seeing in SSDs today. If these tasks spend much of their time doing CPU-bound work, halving the SSD speed might not necessarily correspond to an equivalent drop in boot or app startup times. (Faster SSDs do have measurable benefits, but doubling the SSD speed doesn't nearly correspond to doubling launch times linearly).
In swap-heavy workloads, the performance impacts might be a little more pronounced. It is likely going to be noticeable, but even here, I expect that it isn't going to nearly correspond to a halving of swap performance here either, as the biggest limiting factor for swap is the latency of access and not necessarily the speed of the transfer itself. It can take 100,000 to 300,000 CPU cycles (on average) for a modern SSD to deliver the data for any given requested page to the OS, which is many, many times more than a RAM access (which is closer to 300-400 cycles on the M1). Decreasing the actual transfer rate and cutting it in half might add another 30K to 40K in CPU cycles required to deliver a 16KB page to the OS, but this won't be a severe as cutting the overall performance completely in half (as the latency of access still is a fixed cost that requires hundreds of thousands of cycles in many cases, regardless of transfer speed).
Overall (for most everyday users and use cases) I doubt it is really going to matter all that much whether the SSD is performing at 1300MB/sec in the 256GB models or at ~2600MB/sec in the 512GB models. It's definitely a no brainer to get the faster one if you're able (there are use cases that can take advantage of it, and any performance improvement is welcome for a machine that is expected to be in use for years). But until recently, much of the computer world was still stuck on 540MB/sec SATA SSDs, and those were still blazing fast in comparison to the HDDs that they replaced. Even today, SATA SSDs aren't exactly slow (and can still boot up computers and applications at very fast speeds). So even though it's a little disappointing that Apple cheaped out on the 256GB models, I don't see it necessarily having an earth-shattering impact on the everyday user either (1300MB/sec is still fairly fast in the grand scheme of things). I suppose we will have to wait for more real world benchmarks to see.
