What LLMs fit in 24GB VRAM? (Oct 2026)
26 models on our list fit in 24 GB of graphics memory. The largest by parameter count is Recursive Flint at UD-Q3_K_XL, 19.56 GB (measured weights; total is an estimate). We keep 0.6 GB free on one graphics card, so 23.4 GB is usable. Each total adds conversation memory for 8,192 tokens (or the model's own maximum, when that is shorter) and 0.8 GB for the software. The total is an estimate.
Updated
Nanbeige4.3-33B (33.12B dense) at Q4_K_M needs about 23.3 GB in our estimate, inside the 23.4 GB usable on a 24 GB card. That is the largest dense model at standard quality on this list. Can I run Nanbeige4.3-33B? Qwen3.8-27B needs less memory at its standard file and can step up in quality on the same card. Can I run Qwen3.8-27B? Recursive Flint is the largest mixture-of-experts model that fits 24 GB. Can I run Recursive Flint?
Qwen3.8-27B and Gemma 4 31B both fit one 24 GB card. Qwen3.8-27B vs Gemma 4 31B.
Own a 24 GB GPU? Get a 1-line alert when a new top model fits 23.4 GB usable VRAM
26 models on our catalog fit 24 GB today. Get our weekly 1-line alert when a new open-weight LLM lands inside 23.4 GB at Q4_K_M or when 24 GB+ GPU prices drop >10% — zero spam.
Models that fit
Ranked by parameter count, largest first. The best quant is the highest-quality option that fits in 23.4 GB. A measured download is used when we have one. Otherwise the figure is Q4, Q5, Q8 or FP16 from our estimate. The gigabytes are the weights plus conversation memory for 8,192 tokens, or the model's shorter maximum, plus 0.8 GB for the software. The total is an estimate.
| Rank | AI model | Parameters | Best quant | GB | Label |
|---|---|---|---|---|---|
| 1 | Can I run Recursive Flint? | 39.79B, 2.5B active | UD-Q3_K_XL | 19.56 GB | measured weights; total is an estimate |
| 2 | Can I run Qwen3.6-35B-A3B? | 35.95B (estimate), 3B active | UD-Q4_K_XL | 23.33 GB | measured weights; total is an estimate |
| 3 | Can I run Nanbeige4.3-33B? | 33.12B (estimate) | Q4_K_M | 23.33 GB | measured weights; total is an estimate |
| 4 | Can I run Qwen2.5 32B? | 32.8B (estimate) | Q4_K_M | 22.63 GB | estimate |
| 5 | Can I run Qwen3 32B? | 32.76B (estimate) | Q4_K_M | 22.71 GB | measured weights; total is an estimate |
| 6 | Can I run Gemma 4 31B? | 31.27B (estimate) | Q4_K_M | 20.63 GB | measured weights; total is an estimate |
| 7 | Can I run Qwen3 Coder 30B A3B? | 30.53B (estimate), 3.3B active | Q5_K_M | 23.34 GB | measured weights; total is an estimate |
| 8 | Can I run NVIDIA Nemotron 3.5 Lightning? | 30B (estimate), 3B active | Q4_0 | 19.75 GB | measured weights; total is an estimate |
| 9 | Can I run Muse Glimmer 30B? | 29.6B | KQuant-Dynamic-Q4_K_XL | 20.64 GB | measured weights; total is an estimate |
| 10 | Can I run Qwen3.8-27B? | 27.78B (estimate) | UD-Q6_K | 23.32 GB | measured weights; total is an estimate |
| 11 | Can I run Gemma 3 27B? | 27.43B (estimate) | Q5_K_M | 21.18 GB | measured weights; total is an estimate |
| 12 | Can I run Gemma 2 27B? | 27.2B (estimate) | Q5 (~0.71 bytes/param) | 22.43 GB | estimate |
| 13 | Can I run OrcaCyber Nano 1.0? | 27B | Mix-15.7G | 17.04 GB | measured weights; total is an estimate |
| 14 | Can I run Gemma 4 26B A4B? | 25.2B (estimate), 3.8B active | UD-Q5_K_M | 22.33 GB | measured weights; total is an estimate |
| 15 | Can I run Mistral Small 3.2 24B? | 24.01B (estimate) | Q6_K | 21.49 GB | measured weights; total is an estimate |
| 16 | Can I run gpt-oss-20b? | 21B (estimate), 3.6B active | F16 | 14.79 GB | measured weights; total is an estimate |
| 17 | Can I run DeepSeek R1 Distill Qwen 14B? | 14.77B | Q8_0 | 18.11 GB | measured weights; total is an estimate |
| 18 | Can I run Phi-4 14B? | 14.7B (estimate) | Q8 | 18.21 GB | estimate |
| 19 | Can I run Qwen2.5 14B? | 14.7B (estimate) | Q8 | 18.14 GB | estimate |
| 20 | Can I run Gemma 2 9B? | 9.24B (estimate) | FP16 | 21.39 GB | estimate |
| 21 | Can I run Llama 3.1 8B? | 8.03B (estimate) | FP16 | 17.93 GB | estimate |
| 22 | Can I run NVIDIA Nemotron 3.5 Bolt E4B? | 8B (estimate) | BF16 | 16.12 GB | measured weights; total is an estimate |
| 23 | Can I run Mistral 7B v0.3? | 7.25B (estimate) | FP16 | 16.37 GB | estimate |
| 24 | Can I run Spark-X2.6-4B? | 4.11B (estimate) | BF16 | 9.39 GB | measured weights; total is an estimate |
| 25 | Can I run Llama 3.2 3B? | 3.21B (estimate) | FP16 | 8.16 GB | estimate |
| 26 | Can I run LFM3-1.7B? | 1.73B (estimate) | BF16 | 4.38 GB | measured weights; total is an estimate |
Cards and Macs with 24 GB
The euro price is the figure we store, with the date it was written down. A Mac's gigabytes are shared memory, not graphics memory, so a Mac does not hold the same models as a 24 GB card.
| Machine | Memory | EU price | Date | Label |
|---|---|---|---|---|
| GeForce RTX 3090 24GB | 24 GB graphics memory | €1,246 | Used price, the typical price of second-hand cards that actually sold. Also New price, the lowest in stock in the EU: €2,218, 16 Oct 2026 | |
| Radeon RX 7900 XTX 24GB | 24 GB graphics memory | €1,254 | Used price, the typical price over the last three months. Also New in stock, €3,972 at GPU PRIX on 17 Oct 2026 (unusually high): €3,972, 17 Oct 2026 | |
| GeForce RTX 4090 24GB | 24 GB graphics memory | €2,185 | Used price, the typical price of second-hand cards that actually sold. Also New (lowest listed in the EU, stock not confirmed): €3,790, 21 Oct 2026 |
Running n8n agents on a local model: n8n local LLM hardware requirements. A 24 GB card is the sweet spot. n8n, PostgreSQL and Qdrant need no GPU.
Memory totals are our estimate, including when the weight file was measured. Prices are the euro figures we store, with the date they were written down. How we calculate
Common questions
- What is the largest model that fits in 24 GB of VRAM?
- By total parameter count, Recursive Flint (https://localllmrig.com/can-i-run/flint, 39.79B total, 2.5B active mixture of experts) is the largest at UD-Q3_K_XL, 19.56 GB total (our estimate; 18.49 GB measured weights), followed by Qwen3.6-35B-A3B (https://localllmrig.com/can-i-run/qwen3.6-35b-a3b, 35.95B total, 3B active) at UD-Q4_K_XL, 23.33 GB. Among dense models, where every parameter runs on every word, Nanbeige4.3-33B (https://localllmrig.com/can-i-run/nanbeige-4-3-33b, 33.12B dense, Apache-2.0) is the largest that fits in 24 GB at standard quality: Q4_K_M is 20.79 GB of measured GGUF weights and 23.33 GB total at 8,192 tokens (our estimate), leaving 0.07 GB inside a 24 GB card's 23.4 GB usable ceiling. By comparison, Qwen3.8-27B (https://localllmrig.com/can-i-run/qwen3.8-27b, 27.78B dense) needs 17.80 GB at UD-Q4_K_M and fits all the way up to UD-Q6_K (23.32 GB total) on a 24 GB card. 26 models on our list fit.
- What does the best quant mean for 24 GB?
- The best quant is the highest-quality option that still fits. A measured download is used when we have one at that quality. Otherwise the figure is Q4, Q5, Q8 or FP16 from our estimate. A higher bit count wins. A file that runs only on a Mac is not counted. The gigabytes add conversation memory and the software overhead, so the total is an estimate even when the weight file was measured.
- Which cards and Macs have 24 GB, and what do they cost in the EU?
- GeForce RTX 3090 24GB (24 GB graphics memory): €1,246, Used price, the typical price of second-hand cards that actually sold. Also New price, the lowest in stock in the EU: €2,218, 16 Oct 2026, 24 Oct 2026. Radeon RX 7900 XTX 24GB (24 GB graphics memory): €1,254, Used price, the typical price over the last three months. Also New in stock, €3,972 at GPU PRIX on 17 Oct 2026 (unusually high): €3,972, 17 Oct 2026, 17 Oct 2026. GeForce RTX 4090 24GB (24 GB graphics memory): €2,185, Used price, the typical price of second-hand cards that actually sold. Also New (lowest listed in the EU, stock not confirmed): €3,790, 21 Oct 2026, 16 Oct 2026.
- How much of 24 GB can a model use?
- Our estimate keeps 0.6 GB free on one graphics card, so 23.4 GB is usable. A Mac with 24 GB of shared memory does not give the graphics all of it. The model table is for one graphics card.