Every head-to-head: machine against machine, model against model
Every comparison on this site in one place: 86 machine match-ups and 47 model match-ups, each row carrying the prices, the memory and the speeds the comparison itself opens with. For one machine on its own, start at best buys by usage or the leaderboard.
The short version: none of the 51 machines compared here holds more than 38 of the 39 open models, and 2 of them hold that many. The cheapest that does is the Mac Studio M3 Ultra, 256GB at $7,099. The cheapest machine here at all is the GeForce RTX 3060, 12GB at $329, card only, which holds 11. Both of those prices are the ones those machines launched at, before they were discontinued.
Machine against machine
Five kinds of match-up: one machine per family, the middle of its range by price, against every other; every graphics card against every other card, since a card is bought as a part and a part is what people put against another part; every memory tier of one machine against the others, which is the question left once you have picked the box; each box against the cheapest box built on the same GPU with the same memory, where the whole question is what the dearer one charges on top; and every discontinued machine against the one that replaced it, which is the upgrade question. The two speeds in a row are on the strongest model both machines in it can hold at 32k of context, so they are running the same work.
| Match-up | Price | Memory | Models that fit, of 39 | Speed on a model both hold |
|---|---|---|---|---|
| DGX Spark, 128GB vs GMKtec EVO-X2, 128GB | $4,699 vs $3,500 | 128 GB vs 128 GB | 33 vs 33 | 11 vs 10 tok/s on Qwen3.8 27B, both estimated |
| DGX Spark, 128GB vs MacBook Air M5 (15-inch), 16GB | $4,699 vs $1,499 | 128 GB vs 16 GB | 33 vs 10 | 26 vs 14 tok/s on Gemma 4 12B, both estimated |
| DGX Spark, 128GB vs MacBook Pro M5 Pro (16-inch), 64GB | $4,699 vs $3,999 | 128 GB vs 64 GB | 33 vs 27 | 11 vs 12 tok/s on Qwen3.8 27B, both estimated |
| DGX Spark, 128GB vs Radeon AI PRO R9700, 32GB | $4,699 vs $1,299card only | 128 GB vs 32 GB | 33 vs 27 | 11 vs 30 tok/s on Qwen3.8 27B, both estimated |
| DGX Spark, 128GB vs RTX PRO 6000 Blackwell, 96GB | $4,699 vs $18,000card only | 128 GB vs 96 GB | 33 vs 33 | 11 vs 78 tok/s on Qwen3.8 27B, both estimated |
| Framework Desktop, 128GB vs Beelink GTR9 Pro, 128GB | $3,449 vs $4,349 | 128 GB vs 128 GB | 33 vs 33 | 10 vs 10 tok/s on Qwen3.8 27B, both estimated |
| Framework Desktop, 128GB vs Corsair AI Workstation 300, 128GB | $3,449 vs $4,700 | 128 GB vs 128 GB | 33 vs 33 | 10 vs 10 tok/s on Qwen3.8 27B, both estimated |
| Framework Desktop, 128GB vs GMKtec EVO-X2, 128GB | $3,449 vs $3,500 | 128 GB vs 128 GB | 33 vs 33 | 10 vs 10 tok/s on Qwen3.8 27B, both estimated |
| Framework Desktop, 128GB vs GMKtec EVO-X3, 128GB | $3,449 vs $3,600 | 128 GB vs 128 GB | 33 vs 33 | 10 vs 10 tok/s on Qwen3.8 27B, both estimated |
| Framework Desktop, 128GB vs HP Z2 Mini G1a, 128GB | $3,449 vs $5,544 | 128 GB vs 128 GB | 33 vs 33 | 10 vs 10 tok/s on Qwen3.8 27B, both estimated |
| Framework Desktop, 128GB vs Minisforum MS-S1 Max, 128GB | $3,449 vs $3,799 | 128 GB vs 128 GB | 33 vs 33 | 10 vs 10 tok/s on Qwen3.8 27B, both estimated |
| Framework Desktop, 64GB vs Framework Desktop, 128GB | $1,959 vs $3,449 | 64 GB vs 128 GB | 27 vs 33 | 10 vs 10 tok/s on Qwen3.8 27B, both estimated |
| Framework Desktop, 64GB vs GMKtec EVO-X2, 64GB | $1,959 vs $2,200 | 64 GB vs 64 GB | 27 vs 27 | 10 vs 10 tok/s on Qwen3.8 27B, both estimated |
| GeForce RTX 3090, 24GB vs GeForce RTX 3060, 12GB | $1,499card only vs $329card only | 24 GB vs 12 GB | 24 vs 11 | 88 vs 38 tok/s on Gemma 4 12B, both estimated |
| GeForce RTX 3090, 24GB vs GeForce RTX 4080, 16GB | $1,499card only vs $1,199card only | 24 GB vs 16 GB | 24 vs 13 | 88 vs 74 tok/s on Gemma 4 12B, both estimated |
| GeForce RTX 3090, 24GB vs Radeon AI PRO R9700, 32GB | $1,499card only vs $1,299card only | 24 GB vs 32 GB | 24 vs 27 | 38 vs 30 tok/s on Qwen3.8 27B, both estimated |
| GeForce RTX 4080, 16GB vs GeForce RTX 3060, 12GB | $1,199card only vs $329card only | 16 GB vs 12 GB | 13 vs 11 | 74 vs 38 tok/s on Gemma 4 12B, both estimated |
| GeForce RTX 4090, 24GB vs GeForce RTX 3060, 12GB | $1,599card only vs $329card only | 24 GB vs 12 GB | 24 vs 11 | 102 vs 38 tok/s on Gemma 4 12B, both estimated |
| GeForce RTX 4090, 24GB vs GeForce RTX 3090, 24GB | $1,599card only vs $1,499card only | 24 GB vs 24 GB | 24 vs 24 | 44 vs 38 tok/s on Qwen3.8 27B, both estimated |
| GeForce RTX 4090, 24GB vs GeForce RTX 4080, 16GB | $1,599card only vs $1,199card only | 24 GB vs 16 GB | 24 vs 13 | 102 vs 74 tok/s on Gemma 4 12B, both estimated |
| GeForce RTX 4090, 24GB vs Radeon AI PRO R9700, 32GB | $1,599card only vs $1,299card only | 24 GB vs 32 GB | 24 vs 27 | 44 vs 30 tok/s on Qwen3.8 27B, both estimated |
| GeForce RTX 5090, 32GB vs GeForce RTX 3060, 12GB | $1,999card only vs $329card only | 32 GB vs 12 GB | 27 vs 11 | 155 vs 38 tok/s on Gemma 4 12B, both estimated |
| GeForce RTX 5090, 32GB vs GeForce RTX 3090, 24GB | $1,999card only vs $1,499card only | 32 GB vs 24 GB | 27 vs 24 | 66 vs 38 tok/s on Qwen3.8 27B, both estimated |
| GeForce RTX 5090, 32GB vs GeForce RTX 4080, 16GB | $1,999card only vs $1,199card only | 32 GB vs 16 GB | 27 vs 13 | 155 vs 74 tok/s on Gemma 4 12B, both estimated |
| GeForce RTX 5090, 32GB vs GeForce RTX 4090, 24GB | $1,999card only vs $1,599card only | 32 GB vs 24 GB | 27 vs 24 | 66 vs 44 tok/s on Qwen3.8 27B, both estimated |
| GeForce RTX 5090, 32GB vs Radeon AI PRO R9700, 32GB | $1,999card only vs $1,299card only | 32 GB vs 32 GB | 27 vs 27 | 66 vs 30 tok/s on Qwen3.8 27B, both estimated |
| GMKtec EVO-X2, 128GB vs MacBook Air M5 (15-inch), 16GB | $3,500 vs $1,499 | 128 GB vs 16 GB | 33 vs 10 | 24 vs 14 tok/s on Gemma 4 12B, both estimated |
| GMKtec EVO-X2, 128GB vs MacBook Pro M5 Pro (16-inch), 64GB | $3,500 vs $3,999 | 128 GB vs 64 GB | 33 vs 27 | 10 vs 12 tok/s on Qwen3.8 27B, both estimated |
| GMKtec EVO-X2, 128GB vs Radeon AI PRO R9700, 32GB | $3,500 vs $1,299card only | 128 GB vs 32 GB | 33 vs 27 | 10 vs 30 tok/s on Qwen3.8 27B, both estimated |
| GMKtec EVO-X2, 128GB vs RTX PRO 6000 Blackwell, 96GB | $3,500 vs $18,000card only | 128 GB vs 96 GB | 33 vs 33 | 10 vs 78 tok/s on Qwen3.8 27B, both estimated |
| GMKtec EVO-X2, 64GB vs GMKtec EVO-X2, 128GB | $2,200 vs $3,500 | 64 GB vs 128 GB | 27 vs 33 | 10 vs 10 tok/s on Qwen3.8 27B, both estimated |
| Mac mini M4 Pro, 24GB vs Mac mini M5 Pro, 24GB | $1,399 vs $1,699 | 24 GB vs 24 GB | 13 vs 13 | 26 vs 29 tok/s on Gemma 4 12B, both estimated |
| Mac mini M4 Pro, 48GB vs Mac mini M5 Pro, 48GB | $1,799 vs $2,299 | 48 GB vs 48 GB | 27 vs 27 | 11 vs 12 tok/s on Qwen3.8 27B, both estimated |
| Mac mini M4 Pro, 64GB vs Mac mini M5 Pro, 64GB | $1,999 vs $2,699 | 64 GB vs 64 GB | 27 vs 27 | 11 vs 12 tok/s on Qwen3.8 27B, both estimated |
| Mac mini M4, 16GB vs Mac mini M6, 16GB | $599 vs $899 | 16 GB vs 16 GB | 10 vs 10 | 11 vs 14 tok/s on Gemma 4 12B, both estimated |
| Mac mini M4, 24GB vs Mac mini M6, 24GB | $799 vs $1,099 | 24 GB vs 24 GB | 13 vs 13 | 11 vs 16 tok/s on Gemma 4 12B, both estimated |
| Mac mini M4, 32GB vs Mac mini M6, 32GB | $999 vs $1,299 | 32 GB vs 32 GB | 19 vs 19 | 4.8 vs 6.9 tok/s on Qwen3.8 27B, both estimated |
| Mac mini M5 Pro, 24GB vs DGX Spark, 128GB | $1,699 vs $4,699 | 24 GB vs 128 GB | 13 vs 33 | 29 vs 26 tok/s on Gemma 4 12B, both estimated |
| Mac mini M5 Pro, 24GB vs GMKtec EVO-X2, 128GB | $1,699 vs $3,500 | 24 GB vs 128 GB | 13 vs 33 | 29 vs 24 tok/s on Gemma 4 12B, both estimated |
| Mac mini M5 Pro, 24GB vs Mac mini M5 Pro, 48GB | $1,699 vs $2,299 | 24 GB vs 48 GB | 13 vs 27 | 29 vs 29 tok/s on Gemma 4 12B, both estimated |
| Mac mini M5 Pro, 24GB vs Mac mini M5 Pro, 64GB | $1,699 vs $2,699 | 24 GB vs 64 GB | 13 vs 27 | 29 vs 29 tok/s on Gemma 4 12B, both estimated |
| Mac mini M5 Pro, 24GB vs Mac Studio M5 Max, 128GB | $1,699 vs $5,099 | 24 GB vs 128 GB | 13 vs 33 | 29 vs 58 tok/s on Gemma 4 12B, both estimated |
| Mac mini M5 Pro, 24GB vs MacBook Air M5 (15-inch), 16GB | $1,699 vs $1,499 | 24 GB vs 16 GB | 13 vs 10 | 29 vs 14 tok/s on Gemma 4 12B, both estimated |
| Mac mini M5 Pro, 24GB vs MacBook Pro M5 Pro (16-inch), 64GB | $1,699 vs $3,999 | 24 GB vs 64 GB | 13 vs 27 | 29 vs 29 tok/s on Gemma 4 12B, both estimated |
| Mac mini M5 Pro, 24GB vs Radeon AI PRO R9700, 32GB | $1,699 vs $1,299card only | 24 GB vs 32 GB | 13 vs 27 | 29 vs 70 tok/s on Gemma 4 12B, both estimated |
| Mac mini M5 Pro, 24GB vs RTX PRO 6000 Blackwell, 96GB | $1,699 vs $18,000card only | 24 GB vs 96 GB | 13 vs 33 | 29 vs 182 tok/s on Gemma 4 12B, both estimated |
| Mac mini M5 Pro, 48GB vs Mac mini M5 Pro, 64GB | $2,299 vs $2,699 | 48 GB vs 64 GB | 27 vs 27 | 12 vs 12 tok/s on Qwen3.8 27B, both estimated |
| Mac mini M6, 16GB vs Mac mini M6, 24GB | $899 vs $1,099 | 16 GB vs 24 GB | 10 vs 13 | 14 vs 16 tok/s on Gemma 4 12B, both estimated |
| Mac mini M6, 16GB vs Mac mini M6, 32GB | $899 vs $1,299 | 16 GB vs 32 GB | 10 vs 19 | 14 vs 16 tok/s on Gemma 4 12B, both estimated |
| Mac mini M6, 24GB vs Mac mini M6, 32GB | $1,099 vs $1,299 | 24 GB vs 32 GB | 13 vs 19 | 16 vs 16 tok/s on Gemma 4 12B, both estimated |
| Mac Studio M3 Ultra, 256GB vs Mac Studio M5 Ultra, 256GB | $7,099 vs $10,799 | 256 GB vs 256 GB | 38 vs 38 | 22 vs 32 tok/s on GLM-5.3-Flash, both estimated |
| Mac Studio M3 Ultra, 96GB vs Mac Studio M5 Ultra, 96GB | $3,999 vs $5,499 | 96 GB vs 96 GB | 29 vs 29 | 33 vs 48 tok/s on Qwen3.8 27B, both estimated |
| Mac Studio M4 Max, 128GB vs Mac Studio M5 Max, 128GB | $3,499 vs $5,099 | 128 GB vs 128 GB | 33 vs 33 | 22 vs 25 tok/s on Qwen3.8 27B, both estimated |
| Mac Studio M4 Max, 36GB vs Mac Studio M5 Max, 36GB | $1,999 vs $2,499 | 36 GB vs 36 GB | 27 vs 27 | 17 vs 19 tok/s on Qwen3.8 27B, both estimated |
| Mac Studio M4 Max, 48GB vs Mac Studio M5 Max, 48GB | $2,499 vs $3,099 | 48 GB vs 48 GB | 27 vs 27 | 22 vs 25 tok/s on Qwen3.8 27B, both estimated |
| Mac Studio M4 Max, 64GB vs Mac Studio M5 Max, 64GB | $2,699 vs $3,499 | 64 GB vs 64 GB | 27 vs 27 | 22 vs 25 tok/s on Qwen3.8 27B, both estimated |
| Mac Studio M5 Max, 128GB vs DGX Spark, 128GB | $5,099 vs $4,699 | 128 GB vs 128 GB | 33 vs 33 | 25 vs 11 tok/s on Qwen3.8 27B, both estimated |
| Mac Studio M5 Max, 128GB vs GMKtec EVO-X2, 128GB | $5,099 vs $3,500 | 128 GB vs 128 GB | 33 vs 33 | 25 vs 10 tok/s on Qwen3.8 27B, both estimated |
| Mac Studio M5 Max, 128GB vs MacBook Air M5 (15-inch), 16GB | $5,099 vs $1,499 | 128 GB vs 16 GB | 33 vs 10 | 58 vs 14 tok/s on Gemma 4 12B, both estimated |
| Mac Studio M5 Max, 128GB vs MacBook Pro M5 Pro (16-inch), 64GB | $5,099 vs $3,999 | 128 GB vs 64 GB | 33 vs 27 | 25 vs 12 tok/s on Qwen3.8 27B, both estimated |
| Mac Studio M5 Max, 128GB vs Radeon AI PRO R9700, 32GB | $5,099 vs $1,299card only | 128 GB vs 32 GB | 33 vs 27 | 25 vs 30 tok/s on Qwen3.8 27B, both estimated |
| Mac Studio M5 Max, 128GB vs RTX PRO 6000 Blackwell, 96GB | $5,099 vs $18,000card only | 128 GB vs 96 GB | 33 vs 33 | 25 vs 78 tok/s on Qwen3.8 27B, both estimated |
| Mac Studio M5 Max, 48GB vs Mac Studio M5 Max, 128GB | $3,099 vs $5,099 | 48 GB vs 128 GB | 27 vs 33 | 25 vs 25 tok/s on Qwen3.8 27B, both estimated |
| Mac Studio M5 Max, 48GB vs Mac Studio M5 Max, 64GB | $3,099 vs $3,499 | 48 GB vs 64 GB | 27 vs 27 | 25 vs 25 tok/s on Qwen3.8 27B, both estimated |
| Mac Studio M5 Max, 64GB vs Mac Studio M5 Max, 128GB | $3,499 vs $5,099 | 64 GB vs 128 GB | 27 vs 33 | 25 vs 25 tok/s on Qwen3.8 27B, both estimated |
| MacBook Air M5 (13-inch), 16GB vs MacBook Air M5 (15-inch), 16GB | $1,299 vs $1,499 | 16 GB vs 16 GB | 10 vs 10 | 14 vs 14 tok/s on Gemma 4 12B, both estimated |
| MacBook Air M5 (15-inch), 16GB vs MacBook Pro M5 Pro (16-inch), 64GB | $1,499 vs $3,999 | 16 GB vs 64 GB | 10 vs 27 | 14 vs 29 tok/s on Gemma 4 12B, both estimated |
| MacBook Air M5 (15-inch), 16GB vs Radeon AI PRO R9700, 32GB | $1,499 vs $1,299card only | 16 GB vs 32 GB | 10 vs 27 | 14 vs 70 tok/s on Gemma 4 12B, both estimated |
| MacBook Air M5 (15-inch), 16GB vs RTX PRO 6000 Blackwell, 96GB | $1,499 vs $18,000card only | 16 GB vs 96 GB | 10 vs 33 | 14 vs 182 tok/s on Gemma 4 12B, both estimated |
| MacBook Pro M5 (14-inch), 16GB vs MacBook Pro M5 (14-inch), 32GB | $1,999 vs $2,399 | 16 GB vs 32 GB | 10 vs 19 | 14 vs 14 tok/s on Gemma 4 12B, both estimated |
| MacBook Pro M5 Max (16-inch), 48GB vs MacBook Pro M5 Max (16-inch), 128GB | $4,999 vs $6,999 | 48 GB vs 128 GB | 27 vs 33 | 25 vs 25 tok/s on Qwen3.8 27B, both estimated |
| MacBook Pro M5 Max (16-inch), 48GB vs MacBook Pro M5 Max (16-inch), 64GB | $4,999 vs $5,399 | 48 GB vs 64 GB | 27 vs 27 | 25 vs 25 tok/s on Qwen3.8 27B, both estimated |
| MacBook Pro M5 Max (16-inch), 64GB vs MacBook Pro M5 Max (16-inch), 128GB | $5,399 vs $6,999 | 64 GB vs 128 GB | 27 vs 33 | 25 vs 25 tok/s on Qwen3.8 27B, both estimated |
| MacBook Pro M5 Pro (16-inch), 24GB vs MacBook Pro M5 Pro (16-inch), 48GB | $2,999 vs $3,599 | 24 GB vs 48 GB | 13 vs 27 | 29 vs 29 tok/s on Gemma 4 12B, both estimated |
| MacBook Pro M5 Pro (16-inch), 24GB vs MacBook Pro M5 Pro (16-inch), 64GB | $2,999 vs $3,999 | 24 GB vs 64 GB | 13 vs 27 | 29 vs 29 tok/s on Gemma 4 12B, both estimated |
| MacBook Pro M5 Pro (16-inch), 48GB vs MacBook Pro M5 Pro (16-inch), 64GB | $3,599 vs $3,999 | 48 GB vs 64 GB | 27 vs 27 | 12 vs 12 tok/s on Qwen3.8 27B, both estimated |
| MacBook Pro M5 Pro (16-inch), 64GB vs Radeon AI PRO R9700, 32GB | $3,999 vs $1,299card only | 64 GB vs 32 GB | 27 vs 27 | 12 vs 30 tok/s on Qwen3.8 27B, both estimated |
| MacBook Pro M5 Pro (16-inch), 64GB vs RTX PRO 6000 Blackwell, 96GB | $3,999 vs $18,000card only | 64 GB vs 96 GB | 27 vs 33 | 12 vs 78 tok/s on Qwen3.8 27B, both estimated |
| Radeon AI PRO R9700, 32GB vs GeForce RTX 3060, 12GB | $1,299card only vs $329card only | 32 GB vs 12 GB | 27 vs 11 | 70 vs 38 tok/s on Gemma 4 12B, both estimated |
| Radeon AI PRO R9700, 32GB vs GeForce RTX 4080, 16GB | $1,299card only vs $1,199card only | 32 GB vs 16 GB | 27 vs 13 | 70 vs 74 tok/s on Gemma 4 12B, both estimated |
| RTX PRO 6000 Blackwell, 96GB vs GeForce RTX 3060, 12GB | $18,000card only vs $329card only | 96 GB vs 12 GB | 33 vs 11 | 182 vs 38 tok/s on Gemma 4 12B, both estimated |
| RTX PRO 6000 Blackwell, 96GB vs GeForce RTX 3090, 24GB | $18,000card only vs $1,499card only | 96 GB vs 24 GB | 33 vs 24 | 78 vs 38 tok/s on Qwen3.8 27B, both estimated |
| RTX PRO 6000 Blackwell, 96GB vs GeForce RTX 4080, 16GB | $18,000card only vs $1,199card only | 96 GB vs 16 GB | 33 vs 13 | 182 vs 74 tok/s on Gemma 4 12B, both estimated |
| RTX PRO 6000 Blackwell, 96GB vs GeForce RTX 4090, 24GB | $18,000card only vs $1,599card only | 96 GB vs 24 GB | 33 vs 24 | 78 vs 44 tok/s on Qwen3.8 27B, both estimated |
| RTX PRO 6000 Blackwell, 96GB vs GeForce RTX 5090, 32GB | $18,000card only vs $1,999card only | 96 GB vs 32 GB | 33 vs 27 | 78 vs 66 tok/s on Qwen3.8 27B, both estimated |
| RTX PRO 6000 Blackwell, 96GB vs Radeon AI PRO R9700, 32GB | $18,000card only vs $1,299card only | 96 GB vs 32 GB | 33 vs 27 | 78 vs 30 tok/s on Qwen3.8 27B, both estimated |
Model against model
Each model against the next one down the leaderboard, which is the choice you face once you know what your machine holds. The machine named in a row is the cheapest here that runs both, so the two can be weighed on one computer.
Every figure here is the one the page behind it prints, at the same defaults: 500k tokens a day at 15:1 input to output, 32k of context, $0.17 per kWh, and today's API prices held flat. A speed that says estimated is worked out from memory bandwidth rather than measured. Graphics cards are priced as the card alone, so add the PC around one before comparing it with a complete computer. Scores are the Artificial Analysis Intelligence Index v4.3. Each calculator link opens the machine in its row running the first model named; change any of it once you are there.