If you want to flex on being able to run 397b parameter models at unusably slow tokens/second sure. You can buy a 3090 for $2k, and run QWEN3.5 at 50+ token a

If you want to flex on being able to run 397b parameter models at unusably slow tokens/second sure. You can buy a 3090 for $2k, and run QWEN3.5 at 50+ token a second, and it will do everything you need, especially if you give it enough context.