DeepSeek says both models are more efficient and performant than DeepSeek V3.2 due to architectural improvements, and have almost "closed the gap" with current leading models, both open and closed, on reasoning benchmarks.
It looks like DeepSeek is performing comparably to Opus 4.6 on the benchmarks at least. I haven’t had a chance to drive it much for coding yet, so can’t say for certain how that translates into real world work. But definitely seems promising.
It looks like DeepSeek is performing comparably to Opus 4.6 on the benchmarks at least. I haven’t had a chance to drive it much for coding yet, so can’t say for certain how that translates into real world work. But definitely seems promising.