Yes — with 25.1 GB to spare
LLM-jp 4 33B Thinking at Q4_K_M fits your L40S entirely on the GPU at 8K context, at an estimated 28 tokens per second. There is room for its full 64K window.
Fully on GPU
8K context
Q4_K_M · 18.7 GB
Apache 2.0
Released 14 Aug 2026
New this week
Not in the Ollama library
Japan’s national-institute reasoning model, Japanese and English. A plain dense Llama-style 33B: Q4 is a tight 24 GB fit.
The VRAM budget
weights 18.7 GB
Weights 18.7 GB
KV cache @ 8K 2.00 GB
Runtime overhead 0.6 GB
Free 25.1 GB of 46.4 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| Q8_0 | 32.9 GB | 35.5 GB | 51K | 16 | −0.1% ppl | Long context |
| Q6_K | 25.4 GB | 28.0 GB | 64K | 21 | −0.4% ppl | Long context |
| Q5_K_M | 21.9 GB | 24.5 GB | 64K | 24 | −0.8% ppl | Long context |
| Q4_K_M | 18.7 GB | 21.3 GB | 64K | 28 | −1.9% ppl | Recommended |
| Q3_K_M | 15.1 GB | 17.7 GB | 64K | 35 | −5.4% ppl | Long context |
| Q2_K | 12.9 GB | 15.5 GB | 64K | 40 | −15% ppl | Long context |
Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it.
How to run it
$ llama-server \
-hf llm-jp/llm-jp-4-33b-thinking:Q4_K_M \
-c 8192 -ngl 99
The engine underneath most of the others. Every knob is exposed. More on llama.cpp.
01Download is 18.7 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02Close anything else holding VRAM. A browser with hardware acceleration can sit on 1–2 GB.
03There is room to go to the model's full 64K context on this card.