RTX 4090 vs RTX 5090 vs Mac for local LLMs
Two numbers decide this and the rest is noise. Memory size decides which models you can load at all. Memory bandwidth decides how fast they generate, because producing one token means reading the active weights out of memory once. Everything below comes from the same data as our hardware calculator.
By Evdokiia Petrovskaia, Director, Dubir GroupPrices and specs checked
Short version: buy the 5090 if you want the fastest answers on models up to about 35B. Buy a 128 GB Mac if you want to run the 120B class at all. Buy a used 3090 if you want most of the 4090 for a third of the money. The 4090 in 2026 is the awkward middle: 5090 memory bandwidth at nearly 5090 prices is not what it offers.
The specifications that matter
| RTX 4090 | RTX 5090 | MacBook Pro M5 Max | RTX 3090 (used) | |
|---|---|---|---|---|
| Memory | 24 GB | 32 GB | 128 GB unified | 24 GB |
| Usable by a model | 21.6 GB | 28.8 GB | 96 GB | 21.6 GB |
| Bandwidth | 1008 GB/s | 1792 GB/s | 614 GB/s | 936 GB/s |
| Board power | 450 W | 575 W | 90 W | 350 W |
| Street price | ~$2,900 | ~$4,500 | ~$6,900 | ~$1,250 |
| Models that fit (of 20) | 16 | 16 | 18 | 16 |
Prices are August 2026 street estimates in a year of GDDR7 shortage. The 5090 sells above list, a used 3090 costs more than it did new, and all of it drifts.
Speed, on real models
Estimated single-stream generation at Q4_K_M with 8k of context. These are bandwidth-derived estimates at 60% efficiency, not benchmarks. The ratios are more reliable than the absolute numbers.
| Model | Active params | 4090 | 5090 | M5 Max |
|---|---|---|---|---|
| Gemma 4 12B | 12B | 83 t/s | 148 t/s | 51 t/s |
| Qwen3.6 27B (dense) | 27B | 37 t/s | 66 t/s | 23 t/s |
| gpt-oss 20B (MoE) | 3.6B | 277 t/s | 493 t/s | 169 t/s |
| Qwen3.6 35B-A3B (MoE) | 3B | 333 t/s | 591 t/s | 203 t/s |
| gpt-oss 120B (MoE) | 5.1B | does not fit | does not fit | 119 t/s |
| Laguna S 118B-A8B (MoE) | 8.5B | does not fit | does not fit | 71 t/s |
Two things fall out of that table. The 5090 is a consistent 1.78x on anything both cards run, the exact ratio of their bandwidth. And a dense 27B model is nine times slower than a 35B mixture-of-experts model on the same card, because the MoE reads 3B of weights per token instead of 27B. When people say a bigger model runs faster than a smaller one, this is what they mean, and it is why the 2026 releases are nearly all MoE.
What the Mac buys and what it costs
A 128 GB Mac is the cheapest way to run a 120B model at home. Nothing in the consumer Nvidia line comes close on memory: you would need an RTX PRO 6000 at around $16,000 to match the capacity.
The price is bandwidth. 614 GB/s against 1792 means the M5 Max generates at roughly a third of a 5090's rate on any model both can run. It also draws 90 W against 575 W, which matters if the machine is on all day, and it is a laptop.
The Mac Studio M4 Max at 128 GB is the same 96 GB usable at 546 GB/s for about $4,000, the value pick if you want capacity and do not need it to be portable.
Recommendations
- Fastest answers on models up to 35B: RTX 5090. 1.78x the 4090 for about 55% more money, and the extra 8 GB buys real context headroom.
- Best value: used RTX 3090. Within 7% of the 4090 on generation speed, identical memory, less than half the price.
- Biggest models: 128 GB Mac. It is the only consumer machine here that loads the 120B class, and at 71-119 tokens a second it is genuinely usable rather than a demo.
- Serving several people at once: none of the above on its own . That is a batching question, and the runtime you choose matters more than the card. See Ollama vs LM Studio vs vLLM.
To check a specific machine against a specific model, use the hardware calculator. To decide whether to buy anything at all, the self-hosting breakeven guide is the more important read.
Questions, answered
Is the RTX 5090 worth it over a 4090 for local LLMs?
For speed, yes: 1792 GB/s of memory bandwidth against 1008 means roughly 1.8x the tokens per second on the same model, and the extra 8 GB gives real headroom for longer context. For which models you can run at all, barely: both fit 16 of the 20 models in our catalogue at Q4_K_M, and neither holds a 120B model. You are paying about €1,600 more for speed, not for capability.
Can a Mac run larger models than an RTX 5090?
Yes, and this is the whole argument for Apple hardware. A 128 GB Mac gives a model about 96 GB of usable memory against 28.8 GB on a 32 GB 5090. It runs gpt-oss 120B and Laguna S 118B, which no consumer Nvidia card can load. It runs them at roughly 71-119 tokens a second rather than the several hundred a 5090 manages on smaller models.
How many tokens per second does an RTX 4090 do?
It depends entirely on how many parameters are active per token. A dense 27B model at Q4_K_M runs about 37 tokens a second on a 4090. A 35B mixture-of-experts model with 3B active runs about 333, because only the active experts are read from memory for each token. The architecture matters more than the parameter count.
Is a used RTX 3090 still a good buy?
It is the best value on the list. 936 GB/s against the 4090's 1008 is a 7% difference in generation speed, the same 24 GB of memory, and roughly half to a third of the price. Its weakness is prompt processing and the age of the silicon, not token generation.
Do two GPUs double the speed?
No. Two 24 GB cards give you 48 GB of memory, which lets you run bigger models, but for single-stream generation the layers run in sequence and you get roughly the bandwidth of one card. Two cards buy capacity, not latency.
We build the software behind numbers like these
Dubir Group is a technology company in Paphos, Cyprus. Payroll systems, internal tooling, and AI that runs on your own hardware when the data cannot leave.