Data collected on 12 September 2026
Cinebench 2026 scores and GPU and CPU ranking for 3D rendering
Cinebench 2026 measures your processor and your graphics card with Redshift. Maxon does not publish a ranking of the results. Below, 55 graphics cards and 55 processors are ranked on renders that were actually run, with the number of measurements behind each line.
What this table contains, and what it does not
- None of these figures are Cinebench scores. Maxon distributes Cinebench for free but publishes no results database, so we cannot quote any without copying a third party's table.
- Each line is a median from Blender Open Data, the public database of the Blender Benchmark, which renders a real 3D scene on the tester's machine.
- Median rather than maximum: keeping the maximum would mean publishing, for each model, the most heavily pushed machine ever submitted.
- Only Blender 5.0.0, 5.1.0, 5.1.1, 5.2.0 submissions are kept, that is 24,668 of the 341,466 submissions in the database. Scores do not compare from one major version to another.
- A model only appears from 20 submissions. Below that, a median rests on a handful of machines.
GPU scores: graphics card ranking for rendering
Median Blender Benchmark score per card and per compute backend. The higher the score, the more frames per minute the card renders on the same scene.
| # | Model | Segment | Backend | Median score | Submissions |
|---|---|---|---|---|---|
| 1 | NVIDIA RTX PRO 6000 Blackwell Workstation Edition | Pro workstation | OptiX | 18,024 | 41 |
| 2 | NVIDIA GeForce RTX 5090 | Desktop | OptiX | 16,171 | 485 |
| 3 | NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition | Pro workstation | OptiX | 15,571 | 24 |
| 4 | NVIDIA GeForce RTX 4090 | Desktop | OptiX | 11,534 | 270 |
| 5 | NVIDIA GeForce RTX 5080 | Desktop | OptiX | 9,164 | 818 |
| 6 | NVIDIA GeForce RTX 4080 | Desktop | OptiX | 8,370 | 133 |
| 7 | NVIDIA GeForce RTX 4080 SUPER | Desktop | OptiX | 8,231 | 92 |
| 8 | NVIDIA GeForce RTX 5090 Laptop GPU | Laptop | OptiX | 8,163 | 101 |
| 9 | Apple M5 Max (GPU - 40 cores) | Apple Silicon | Metal | 8,107 | 211 |
| 10 | NVIDIA GeForce RTX 5070 Ti | Desktop | OptiX | 7,705 | 592 |
| 11 | NVIDIA GeForce RTX 4070 Ti SUPER | Desktop | OptiX | 7,052 | 107 |
| 12 | NVIDIA RTX PRO 4000 Blackwell | Pro workstation | OptiX | 6,862 | 21 |
| 13 | NVIDIA GeForce RTX 4090 Laptop GPU | Laptop | OptiX | 6,686 | 94 |
| 14 | NVIDIA GeForce RTX 5080 Laptop GPU | Laptop | OptiX | 6,656 | 216 |
| 15 | NVIDIA GeForce RTX 4070 Ti | Desktop | OptiX | 6,438 | 77 |
| 16 | Apple M5 Max (GPU - 32 cores) | Apple Silicon | Metal | 6,432 | 36 |
| 17 | NVIDIA GeForce RTX 4070 SUPER | Desktop | OptiX | 6,082 | 261 |
| 18 | NVIDIA GeForce RTX 5070 | Desktop | OptiX | 5,879 | 472 |
| 19 | Apple M4 Max (GPU - 40 cores) | Apple Silicon | Metal | 5,841 | 60 |
| 20 | NVIDIA GeForce RTX 4080 Laptop GPU | Laptop | OptiX | 5,309 | 74 |
| 21 | NVIDIA GeForce RTX 3090 Ti | Desktop | OptiX | 5,251 | 29 |
| 22 | NVIDIA GeForce RTX 4070 | Desktop | OptiX | 5,217 | 166 |
| 23 | NVIDIA GeForce RTX 3090 | Desktop | OptiX | 5,131 | 231 |
| 24 | NVIDIA GeForce RTX 5070 Ti Laptop GPU | Laptop | OptiX | 5,071 | 137 |
| 25 | NVIDIA GeForce RTX 3080 Ti | Desktop | OptiX | 5,029 | 139 |
| 26 | AMD Radeon RX 7900 XTX | Desktop | HIP | 4,894 | 194 |
| 27 | Apple M4 Max (GPU - 32 cores) | Apple Silicon | Metal | 4,826 | 29 |
| 28 | Apple M5 Pro (GPU - 20 cores) | Apple Silicon | Metal | 4,300 | 97 |
| 29 | NVIDIA GeForce RTX 3080 | Desktop | OptiX | 4,289 | 248 |
| 30 | NVIDIA GeForce RTX 5060 Ti | Desktop | OptiX | 4,249 | 364 |
| 31 | NVIDIA GeForce RTX 5070 Laptop GPU | Laptop | OptiX | 3,997 | 154 |
| 32 | AMD Radeon RX 7900 XT | Desktop | HIP | 3,811 | 46 |
| 33 | NVIDIA GeForce RTX 4060 Ti | Desktop | OptiX | 3,805 | 138 |
| 34 | Intel Arc Pro B70 Graphics | Desktop | oneAPI | 3,802 | 23 |
| 35 | NVIDIA RTX 3500 Ada Generation Laptop GPU | Laptop | OptiX | 3,759 | 20 |
| 36 | Apple M5 Pro (GPU - 16 cores) | Apple Silicon | Metal | 3,746 | 47 |
| 37 | NVIDIA GeForce RTX 5060 | Desktop | OptiX | 3,681 | 210 |
| 38 | NVIDIA GeForce RTX 4070 Laptop GPU | Laptop | OptiX | 3,535 | 91 |
| 39 | AMD Radeon RX 9070 XT | Desktop | HIP | 3,443 | 621 |
| 40 | AMD Radeon RX 7900 GRE | Desktop | HIP | 3,387 | 42 |
| 41 | NVIDIA GeForce RTX 5060 Laptop GPU | Laptop | OptiX | 3,296 | 283 |
| 42 | NVIDIA GeForce RTX 3070 Ti | Desktop | OptiX | 3,282 | 64 |
| 43 | AMD Radeon AI PRO R9700 | Desktop | HIP | 3,279 | 21 |
| 44 | NVIDIA GeForce RTX 3070 | Desktop | OptiX | 3,051 | 181 |
| 45 | NVIDIA GeForce RTX 4060 | Desktop | OptiX | 3,048 | 172 |
| 46 | AMD Radeon RX 9070 | Desktop | HIP | 3,025 | 46 |
| 47 | NVIDIA GeForce RTX 4060 Laptop GPU | Laptop | OptiX | 2,983 | 245 |
| 48 | NVIDIA GeForce RTX 3080 Laptop GPU | Laptop | OptiX | 2,971 | 51 |
| 49 | Apple M4 Pro (GPU - 20 cores) | Apple Silicon | Metal | 2,949 | 42 |
| 50 | AMD Radeon RX 7800 XT | Desktop | HIP | 2,844 | 47 |
| 51 | NVIDIA GeForce RTX 5050 | Desktop | OptiX | 2,777 | 53 |
| 52 | Intel Arc B580 Graphics | Desktop | oneAPI | 2,772 | 66 |
| 53 | AMD Radeon RX 6800 XT | Desktop | HIP | 2,753 | 41 |
| 54 | NVIDIA GeForce RTX 2080 Ti | Desktop | OptiX | 2,749 | 31 |
| 55 | NVIDIA GeForce RTX 5050 Laptop GPU | Laptop | OptiX | 2,647 | 130 |
CPU scores: processor ranking for rendering
Same scene and same points scale as the graphics card table, which makes the two directly comparable.
| # | Model | Segment | Compute | Median score | Submissions |
|---|---|---|---|---|---|
| 1 | AMD Ryzen 9 9950X3D2 16-Core Processor | Desktop | CPU | 737 | 107 |
| 2 | AMD Ryzen 9 9950X3D 16-Core Processor | Desktop | CPU | 703 | 435 |
| 3 | AMD Ryzen 9 9950X 16-Core Processor | Desktop | CPU | 671 | 323 |
| 4 | Intel Core Ultra 7 270K Plus | Desktop | CPU | 634 | 168 |
| 5 | AMD Ryzen 9 9955HX 16-Core Processor | Laptop | CPU | 631 | 31 |
| 6 | Intel Core Ultra 9 285K | Desktop | CPU | 596 | 150 |
| 7 | AMD Ryzen 9 7950X3D 16-Core Processor | Desktop | CPU | 588 | 93 |
| 8 | AMD Ryzen 9 7950X 16-Core Processor | Desktop | CPU | 579 | 78 |
| 9 | Apple M5 Max | Apple Silicon | CPU | 575 | 147 |
| 10 | Intel Core Ultra 9 290HX Plus | Laptop | CPU | 546 | 26 |
| 11 | AMD Ryzen 9 9900X3D 12-Core Processor | Desktop | CPU | 540 | 23 |
| 12 | Intel Core i9-14900KF | Desktop | CPU | 537 | 27 |
| 13 | Apple M5 Pro | Apple Silicon | CPU | 533 | 139 |
| 14 | Intel Core i9-14900K | Desktop | CPU | 530 | 155 |
| 15 | AMD Ryzen 9 9900X 12-Core Processor | Desktop | CPU | 522 | 184 |
| 16 | 13th Gen Intel Core i9-13900K | Desktop | CPU | 517 | 37 |
| 17 | Intel Core Ultra 7 265K | Desktop | CPU | 504 | 108 |
| 18 | Intel Core Ultra 7 265KF | Desktop | CPU | 502 | 22 |
| 19 | Intel Core Ultra 9 275HX | Laptop | CPU | 486 | 119 |
| 20 | Intel Core Ultra 5 250KF Plus | Desktop | CPU | 478 | 20 |
| 21 | AMD Ryzen 9 8940HX | Laptop | CPU | 477 | 29 |
| 22 | Intel Core Ultra 5 250K Plus | Desktop | CPU | 467 | 42 |
| 23 | Intel Core i7-14700K | Desktop | CPU | 453 | 53 |
| 24 | Apple M4 Max | Apple Silicon | CPU | 451 | 63 |
| 25 | Intel Core i7-14700KF | Desktop | CPU | 436 | 72 |
| 26 | AMD RYZEN AI MAX+ 395 | Desktop | CPU | 423 | 50 |
| 27 | INTEL XEON PLATINUM 8581C CPU @ 2.30GHz | Pro workstation | CPU | 423 | 30 |
| 28 | AMD Ryzen 9 7900X 12-Core Processor | Desktop | CPU | 421 | 233 |
| 29 | AMD Ryzen 9 5900XT 16-Core Processor | Desktop | CPU | 419 | 79 |
| 30 | AMD Ryzen 9 5950X 16-Core Processor | Desktop | CPU | 409 | 276 |
| 31 | AMD Ryzen 9 7900 12-Core Processor | Desktop | CPU | 402 | 27 |
| 32 | 13th Gen Intel Core i7-13700K | Desktop | CPU | 398 | 34 |
| 33 | Intel Core Ultra 7 255HX | Laptop | CPU | 395 | 28 |
| 34 | 13th Gen Intel Core i7-13700KF | Desktop | CPU | 387 | 29 |
| 35 | 13th Gen Intel Core i9-13900HX | Laptop | CPU | 382 | 32 |
| 36 | 13th Gen Intel Core i9-13980HX | Laptop | CPU | 379 | 47 |
| 37 | AMD Ryzen 7 9850X3D 8-Core Processor | Desktop | CPU | 378 | 93 |
| 38 | Apple M3 Max | Apple Silicon | CPU | 373 | 29 |
| 39 | AMD Ryzen 7 9800X3D 8-Core Processor | Desktop | CPU | 369 | 409 |
| 40 | Intel Core i9-12900K | Desktop | CPU | 363 | 22 |
| 41 | Intel Core i9-14900HX | Laptop | CPU | 359 | 29 |
| 42 | Apple M4 Pro | Apple Silicon | CPU | 358 | 72 |
| 43 | Intel Core i7-14700HX | Laptop | CPU | 333 | 28 |
| 44 | Intel Core i5-14600KF | Desktop | CPU | 332 | 38 |
| 45 | Intel Core i5-14600K | Desktop | CPU | 327 | 41 |
| 46 | AMD Ryzen 7 9700X 8-Core Processor | Desktop | CPU | 327 | 216 |
| 47 | 13th Gen Intel Core i5-13600KF | Desktop | CPU | 327 | 25 |
| 48 | AMD Ryzen 9 5900X 12-Core Processor | Desktop | CPU | 324 | 133 |
| 49 | AMD Ryzen AI 9 HX 370 | Laptop | CPU | 321 | 33 |
| 50 | 13th Gen Intel Core i5-13600K | Desktop | CPU | 314 | 24 |
| 51 | Intel Core i7-12700K | Desktop | CPU | 309 | 40 |
| 52 | Intel Core i7-14650HX | Laptop | CPU | 300 | 35 |
| 53 | AMD Ryzen 7 7700X 8-Core Processor | Desktop | CPU | 297 | 57 |
| 54 | AMD Ryzen 7 7800X3D 8-Core Processor | Desktop | CPU | 294 | 184 |
| 55 | AMD Ryzen 7 7700 8-Core Processor | Desktop | CPU | 287 | 55 |
How to read these scores
The score is proportional to the number of frames rendered per minute on the test scene. A ratio therefore reads directly: a machine rated twice as high finishes the same scene in half the time, with identical settings.
The first gap in the data of 12 September 2026 is the one between the processor and the graphics card. At the top of the ranking, the NVIDIA RTX PRO 6000 Blackwell Workstation Edition scores 18,024 points with OptiX, while the AMD Ryzen 9 9950X3D2 16-Core Processor scores 737, a ratio of 24 to 1. This is why a processor score, Cinebench R23 included, says almost nothing about how long a Cycles render on a graphics card will take. If you are choosing a machine from these gaps, the guide best PC for 3D rendering reviews the graphics card, the VRAM and the processor.
The backend matters as much as the chip. The same NVIDIA card is measured with OptiX and with CUDA, a Radeon with HIP, a Mac with Metal, and the scores differ. The column states it for each line.
These medians describe the testers' machines. What our own machines measure on customer scenes, Cycles and EEVEE, is published on the Blender benchmark page.
Cinebench 2026 and Cinebench R23: what each version measures
According to Maxon's product page, consulted on 12 September 2026, Cinebench 2026 relies on Redshift, the default render engine of Cinema 4D, to evaluate the processor and the graphics card. The same test scene is used for both measurements. The tool covers x86/64 on Windows and macOS, Apple Silicon on macOS and Arm64 processors on Windows. It is free.
This is the change to remember compared with Cinebench R23, which only rated the processor. An R23 score and a 2026 score do not measure the same thing and do not convert into each other: the scene and the engine have changed. If you compare two machines, compare two scores from the same version, exactly as the table above is limited to Blender 5.0.0, 5.1.0, 5.1.1, 5.2.0.
For more on the versions and on reading a score, our dedicated guide covers the question: what is a good Cinebench score for 3D rendering (in French).
Your machine is at the bottom of the table: what does rendering elsewhere cost?
Changing your graphics card to climb this ranking costs thousands of euros. Renting GPU time from a Blender render farm is priced by the hour. TURNA sends your Blender scenes, Cycles and Eevee, from an add-on and splits them across several machines in parallel. Machines are sought in Europe first, and you are billed for the time actually used, with no rounding to the hour.
Prices exclude VAT. For individuals in France, 20% VAT is added, which makes €5.88 per hour on Standard. The total including VAT is always shown before payment.
Frequently asked questions
Does Cinebench 2026 test the graphics card?
Yes. Maxon states that Cinebench 2026 relies on Redshift, the Cinema 4D render engine, to evaluate the processor and the graphics card, with the same test scene for both. This is the main difference from Cinebench R23, which only measured the processor.
Where can I find an official ranking of Cinebench 2026 scores?
Maxon distributes Cinebench for free but does not publish a results database you can browse online. The rankings found on the web are third-party records, rarely dated and rarely reproducible. So instead of copying Cinebench scores, this page publishes the medians of Blender Open Data, the public database of the Blender Benchmark, where every submission can be checked at the source.
Is a Cinebench R23 score comparable to a Cinebench 2026 score?
No. The scene and the engine changed between the two versions, Cinebench 2026 being based on Redshift. Two scores from different versions do not measure the same thing and do not convert into each other. The same principle applies to the figures on this page, which only keep Blender 5.0.0, 5.1.0, 5.1.1, 5.2.0 submissions.
What score should I aim for in 3D rendering?
Rather than an arbitrary threshold, compare your machine with those in the table: the score is proportional to the number of frames rendered per minute on the same scene. A machine rated twice as high renders the same scene in half the time. The TURNA render time calculator applies this ratio to your frame count to estimate a duration.
Sources and method
- Scores: Blender Open Data API, collected on 12 September 2026. Median per device and backend pair, on Blender 5.0.0, 5.1.0, 5.1.1, 5.2.0 submissions, from 20 submissions upwards. The collection can be reproduced with
scripts/generer_scores_opendata.py. - Description of Cinebench 2026: maxon.net, consulted on 12 September 2026.
- TURNA pricing: grid published on the Pricing page.
- Blender Open Data is a service of the Blender Foundation. TURNA has no part in it and does not control it. Cinebench and Redshift are trademarks of Maxon.
