Can the NVIDIA GeForce RTX 4090 run Kimi-Linear-48B-A3B-Instruct?

Kimi-Linear-48B-A3B-Instruct · NVIDIA GeForce RTX 4090 · 8,192 ctx · whole model residentQ2_K sizereconstructed sizeSpeedestimate

IT FITS

Yes. 1 of 9 formats evaluated fit in 24 GB, at 8,192 tokens. No published file exists for this pair, so every size is rebuilt from the pinned architecture. The best quality that fits is Q2_K, needing 18.8 GB and running at an estimated 523 tokens per second.

Fitting in memory is not the same as loading. Whether the runtime and version you have supports this architecture and format on this machine has not been tested here.

Find the NVIDIA GeForce RTX 4090: Amazon ↗ · eBay (new and used) ↗Store links may pay us a commission. They never decide which card is suggested — the memory arithmetic does.

1/9formats that fit

24 GBdevice memory

18.8 GBneeded at best quality

~523tokens per second, estimatedFaster than you read

1008GB/s bandwidth

Run it in the cloud

Run Kimi-Linear-48B-A3B-Instruct properly, for cents an hour

The NVIDIA GeForce RTX 4090 only holds it at Q2_K, at an estimated ~523 tokens per second. A rented card runs a better format at full speed, billed by the second.

Your card Or keep Q4_K_M on your own card: with about 7 GB of it in system RAM, the NVIDIA GeForce RTX 4090 runs it at ~115 tokens per second (faster than you read) — free, if you have the RAM. A rented card below holds all of it on the GPU. Size the offload →

Prices · 18:30 UTC, 22 Sept

Cheapest48 GB

4× RTX 3060

$0.21per hour

Speed: ~443 tok/sFaster than you read
Per million tokens: $0.13

Vast.ai · Marketplace · 99.2% reliable

Rent on Vast.ai

Lowest price per hour

Recommended48 GB

2× RTX 3090

$0.24per hour

Speed: ~642 tok/sFaster than you read
Per million tokens: $0.10

Vast.ai · Marketplace · 99.5% reliable

Rent on Vast.ai

Lowest cost per token at 20+ tok/s

1.9× faster80 GB

1× H100 SXM

$1.74per hour

Speed: ~1218 tok/sFaster than you read
Per million tokens: $0.40

Vast.ai · Marketplace · 98.7% reliable

Rent on Vast.ai

or RunPod $3.49/h

Compare all 23 rentable configurations →

First time renting a GPU? How it works, in four steps

Create an account and add credit. Both providers are prepaid: $10 is enough to try any card on this page for hours. A new RunPod account that arrives through a referral link gets a one-time $5 credit when it first adds $10.
Pick the card and the count shown here, and a template: llama.cpp or vLLM for an OpenAI-compatible API, or one with a web chat if you only want to talk to the model. “How to launch it” below gives the exact image and command.
Wait for the download. The weights are fetched on the machine; tens of gigabytes take a few minutes on a data-centre connection. You pay by the second from the moment it starts.
Stop it when you are done. A running machine bills even when idle. A stopped one costs nothing per hour, though its disk is usually still billed until you delete it.

How to launch it

No Q4_K_M file of Kimi-Linear-48B-A3B-Instruct is catalogued here, so there is no exact command to vouch for. Start the machine from the llama.cpp server image ghcr.io/ggml-org/llama.cpp:server-cuda and point -hf at a Q4_K_M build from the publisher or a quantizer you trust — search Hugging Face for one. Check its file size against the memory figure on this page before you rent.

Every machine holds the whole model at Q4_K_M and 8,192 tokens of context, no offload. Speeds are this site’s single-stream decode estimates; prices are what each provider’s own API quoted, on-demand, for the whole machine. Vast hosts below 98% measured reliability are left out.

Referral links Vast.ai, RunPod and Novita pay us a share of what you spend if you sign up through these buttons. It costs you nothing, and it never decides an order or a recommendation: both are computed from the live price and the speed, and options that pay us nothing are listed and recommended on the same terms. How we rank

Runs well on your NVIDIA GeForce RTX 4090 instead

The models nearest to Kimi-Linear-48B-A3B-Instruct — same lab first, then closest in size — that the NVIDIA GeForce RTX 4090 holds whole at the standard format and runs at a usable speed. Nearest in size is not the same as equally good; compare them on the task you care about.

Qwen-AgentWorld-35B-A3BQ4_K_M · 21.8 GB · ~370 tok/s
Qwen3.6-35B-A3BQ4_K_M · 23.1 GB · ~370 tok/s
Qwen3.5-35B-A3BQ4_K_M · 23.1 GB · ~370 tok/s
granite-4.2-30bQ4_K_M · 20.7 GB · ~35 tok/s

How long a conversation. At Q2_K, this pair holds 658,035 tokens of context — roughly 493,526 words — needing 24.0 GB. Past that the NVIDIA GeForce RTX 4090 runs out of memory, not the model out of context. This is the wall a long chat hits after it has already loaded fine. The nearest round setting below it is 524,288.

Every format evaluated

No published file exists for this pair, so every size is rebuilt from the pinned architecture. The Basis column says which is which for each row. Decode and first-token figures are calibrated estimates, not runs on this card.

Format	Needs	Of 24 GB	Fits	Decode	First token	Basis
FP16	98.4 GB	410%	short by 74.4 GB	does not run	—	reconstructed size
Q8_0	52.7 GB	220%	short by 28.7 GB	does not run	—	reconstructed size
Q6_K	40.9 GB	171%	short by 16.9 GB	does not run	—	reconstructed size
Q5_K_M	35.5 GB	148%	short by 11.5 GB	does not run	—	reconstructed size
Q5_0	34.5 GB	144%	short by 10.5 GB	does not run	—	reconstructed size
Q4_K_M	30.4 GB	127%	short by 6.37 GB	does not run	—	reconstructed size
Q4_0	28.5 GB	119%	short by 4.48 GB	does not run	—	reconstructed size
Q3_K_M	25.1 GB	104%	short by 1.08 GB	does not run	—	reconstructed size
Q2_K	18.8 GB	78%	yes	~523 tok/s	~2.8 s	reconstructed size

Sized at 8,192 tokens of context with the whole model resident — weights, the KV cache and the runtime reserve, offload off. Speed is only quoted for a format that fits: a rate for a configuration that cannot load is not a fact about anything.

Why the first-token figure is the same on every row

First token is modelled from the arithmetic the prompt requires, and that count does not change with the weight format — which is why it reads the same on every row. Real prefill does vary by format, because a quantized matmul is a different kernel; this model does not capture that, and the figure should be read as an order of magnitude rather than a ranking between formats.

What the labels mean

published data: Read from a published source — a file's byte count, a model's configuration or a manufacturer's specification — or exact arithmetic on such values. Not a measurement on a machine.
reconstructed size: Weight size reconstructed from the pinned architecture, because no published file exists.
size range: Only a lower and an upper bound are claimed for this weight size.
estimate: Calculated from sourced inputs by a stated method; an estimate, not a measurement.

Change anything

This page fixes the context at 8,192 tokens and one device. Batch, concurrent users, KV-cache format, clusters and rental cost are all in the calculator, already set to this pairing.

Open in the calculator →Everything the NVIDIA GeForce RTX 4090 runs Best models for 24 GB

Where these numbers come from

Kimi-Linear-48B-A3B-Instruct’s architecture is read from its publisher’s own config.json at a pinned revision, and the NVIDIA GeForce RTX 4090’s 24 GB and 1008 GB/s come from the manufacturer’s specification. How far each figure can be trusted is published, per format and worst case included, on the accuracy scorecard.

Building this into your own product? A free API key returns exactly these figures, and the widget puts this answer on a product page with one script tag.

Use this answer in your own product

As JSON, from the API

curl -s https://llmbottleneck.com/v1/analyze \
  -H "Authorization: Bearer $LLMB_KEY" -H "content-type: application/json" \
  -d '{"model":"moonshotai-kimi-linear-48b-a3b-instruct","quantization":"Q2_K","context":8192,"hardware":"nvidia-geforce-rtx-4090"}'

Same engine, same evidence, every field sourced. Free key in one step →

On your page, as a widget

<script src="https://llmbottleneck.com/widget.js"
  data-model="moonshotai-kimi-linear-48b-a3b-instruct" data-quantization="Q2_K"
  data-hardware="nvidia-geforce-rtx-4090" data-context="8192"></script>

No key needed. Unbranded, with your own buy button, on Pro and Business →