Cinebench is the go-to CPU benchmark for creative and productivity work, but a common mistake trips people up: comparing a Cinebench R23 score to a Cinebench R24 (Cinebench 2024) score as if they're the same scale. They aren't. The two versions use different underlying render engines and workloads, so their numbers don't translate, and cross-referencing them leads to wrong conclusions about CPU performance. This article explains what each version measures and why you must compare within a version, not across.
It relates to our guides on CPU cores vs threads and GPU architecture for rendering, and the benchmark-reading hub.
What Changed Between R23 and R24
- Different render engine: R23 was based on an older rendering technology, while R24 (Cinebench 2024) moved to a newer engine derived from Maxon's Redshift, with a heavier, more modern workload. The work being timed is genuinely different.
- A GPU test added: R24 adds a GPU benchmark alongside the CPU test, reflecting how modern rendering uses the GPU — something R23's CPU-focused test didn't cover.
- Different score scale: because the workload changed, R24 scores live on a different scale from R23. A number that looks lower isn't necessarily worse — it's a different measurement.
Why You Can't Cross-Reference
Since the engine and workload differ, an R23 score and an R24 score are not points on the same scale. Comparing them tells you nothing reliable about relative performance. The only valid comparisons are within a single version: R23 score vs R23 score, or R24 vs R24. If a review quotes R23 and you find another quoting R24, you can't combine them — you need both CPUs measured in the same version.
Which Version to Use
R24 better reflects modern rendering (and adds the GPU test), so it's the more representative choice going forward — but R23 still has a vast library of existing results, which is useful when comparing against older CPUs only measured in R23. Use whichever version both CPUs you're comparing were tested in, and never mix the two. Cinebench is a multi-threaded render workload, so it favours core count — keep that in mind when reading it against single-thread-bound applications.
Frequently Asked Questions
Can I compare a Cinebench R23 score to an R24 score? No — they use different render engines and workloads, so the scores are on different scales and don't translate. Only compare within a single version: R23 to R23, or R24 to R24. Mixing them produces misleading conclusions.
What's the difference between R23 and R24? R24 (Cinebench 2024) uses a newer render engine derived from Redshift with a heavier modern workload, and adds a GPU benchmark alongside the CPU test. R23 used an older engine and was CPU-focused. The work being timed is genuinely different.
Which Cinebench version should I use? R24 is more representative of modern rendering and adds a GPU test, so prefer it going forward — but use whichever version both CPUs you're comparing were tested in. R23's larger existing result library is handy for comparing against older CPUs.
The One Thing to Remember
Cinebench R23 and R24 measure different things with different engines, so their scores sit on different scales and must never be cross-referenced — compare only within a single version. R24 better reflects modern rendering and adds a GPU test; R23 has more legacy results. Cinebench is a multi-threaded render workload that rewards cores, so read it in that light, and always match the version across the CPUs you compare.
Choosing a CPU for rendering or productivity? Configure a workstation online → or talk to our team → and we'll spec it to your actual workload, not a mismatched benchmark number.