Can the NVIDIA GeForce RTX 3090 run Aya 23 35B?

Yes, it runs well (offloads to RAM)

Runs well (offloads to RAM). At Q2_K it needs about 26.5 GB of 24.0 GB usable, with an estimated speed of 16.1 tok/s.

Best quant: Q2_K
Generation speed: 16.1 tok/s
Prompt processing: 640 tok/s
First token (1K prompt): 1.6 s
Largest safe context: 6K

Best app to run it on the NVIDIA GeForce RTX 3090

On Windows

LM Studio

Desktop app · runs on llama.cpp, MLX

Splits the model between GPU memory and system RAM, which this setup needs.
Point-and-click: search, download and chat in one window.

Open Aya 23 35B in LM Studio

Ollama Splits the model between GPU memory and system RAM, which this setup needs.
llama.cpp Splits the model between GPU memory and system RAM, which this setup needs.

On Linux

LM Studio

Desktop app · runs on llama.cpp, MLX

Splits the model between GPU memory and system RAM, which this setup needs.
Point-and-click: search, download and chat in one window.

Open Aya 23 35B in LM Studio

Ollama Splits the model between GPU memory and system RAM, which this setup needs.
llama.cpp Splits the model between GPU memory and system RAM, which this setup needs.

How we choose: the files published for this model, your operating system and GPU, whether it fits fully in memory, and published speed tests (MLX vs llama.cpp on Apple silicon, vLLM vs llama.cpp vs Ollama on NVIDIA). Details.

Download Aya 23 35B

Recommended for the NVIDIA GeForce RTX 3090: Q2_K. Every other quantization has its own download link in the table below.

Quickest · Ollama

One command downloads a ready-to-run version and starts a chat.

Get Ollama at ollama.com/download (opens in a new tab) · model page (opens in a new tab)

ollama run aya:35b-23-q2_K

Quantized files · GGUF

Works in LM Studio, llama.cpp, Jan and Ollama. No sign-in needed.

Download the Q2_K file (opens in a new tab)

huggingface.co/bartowski/aya-23-35B-GGUF (opens in a new tab)

ollama run hf.co/bartowski/aya-23-35B-GGUF:Q2_K

Original weights · Hugging Face

Full-precision files from Cohere, for fine-tuning, vLLM, Transformers.

huggingface.co/CohereLabs/aya-23-35B (opens in a new tab)

Free Hugging Face account needed: accept the license on the model page and the download starts right away.

Links checked 3 days ago against Hugging Face, Ollama and GitHub.

How it’s rated

Quality 35Basic quality (D) · 35 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking.: How capable the model itself is at Q2_K, from its editorial quality tier scaled by the quantization or precision. It’s the same on any hardware that runs this variant. Scale: A+ 85+ · A 70+ · B 55+ · C 40+ · D under 40.
Runs Runs welloffloads to RAM: How well it runs on the NVIDIA GeForce RTX 3090: speed, capped by how well it fits in memory. Runs great: 30+ tokens/s (or near the image, video or decision speed target) · Runs well: 10+ tokens/s · Slow: 3+ tokens/s · Very slow: under 3 · Won’t run: doesn’t fit even with offload. A tight fit or RAM offload caps it at Runs well; heavy offload caps it at Slow.

Match score

To order lists, each result also gets a match score out of 100 from four parts. It favors the most capable model that still runs well.

Model qualityEditorial quality tier of the model, scaled by the quantization or precision used.	15.8 / 45
Memory fitFull marks when everything sits in GPU memory; fewer when part spills to system RAM.	10.0 / 25
SpeedGeneration speed against a 40 tokens/s target for LLMs, or render time against a target for image and video.	9.5 / 15
Memory useRewards using the memory you have, so a bigger, better variant beats a tiny one that leaves most memory idle.	15.0 / 15
Match score	50.3 / 100

The match score orders results; it is not shown as a grade. Quality and Runs are the two ratings in the tables.

Every quantization on the NVIDIA GeForce RTX 3090

Quant	Memory needed	Available	Speed	Runs	Quality	Download
`Q2_K`	26.5 GB	24.0 GB	16.1 tok/s	Runs welloffloads to RAM	35Basic quality (D) · 35 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q3_K_M`	29.7 GB	24.0 GB	8.0 tok/s	Slowheavy RAM offload	42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 44/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q4_K_M`	33.9 GB	24.0 GB	4.7 tok/s	Slowheavy RAM offload	47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 45/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q5_K_M`	37.4 GB	24.0 GB	3.4 tok/s	Slowheavy RAM offload	48Fair quality (C) · 48 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 45/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q6_K`	41.2 GB	24.0 GB	2.6 tok/s	Very slowheavy RAM offload	49Fair quality (C) · 49 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 45/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q8_0`	49.6 GB	24.0 GB	—	Won't run	50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`F16`	82.5 GB	24.0 GB	—	Won't run	50Fair quality (C) · 50 of 100: how capable the model is at F16, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking.	GGUF · folder (opens in a new tab) Original weights (opens in a new tab)

Memory figures assume 8K of context. Longer conversations or documents grow the KV cache.

Other hardware that runs it

49Fair quality (C) · 49 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA RTX PRO 6000 Blackwell Q6_K · 43.6 tok/s · Runs great
42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 5090 Q3_K_M · 81.3 tok/s · Runs well (tight fit)
47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. Apple M4 Max (64 GB) Q4_K_M · 16.5 tok/s · Runs well
47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. Apple M1 Max (64 GB) Q4_K_M · 12.1 tok/s · Runs well
47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. Apple M4 Max (128 GB) Q4_K_M · 16.5 tok/s · Runs well
42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. Apple M4 Pro (48 GB) Q3_K_M · 10.7 tok/s · Runs well (tight fit)

Quick answers

How much memory does Aya 23 35B need on the RTX 3090 24GB?

At Q2_K it needs about 26.5 GB. The RTX 3090 24GB has 24.0 GB usable.

How fast is Aya 23 35B on the RTX 3090 24GB?

About 16.1 tok/s at Q2_K, an estimate from the hardware’s memory bandwidth and the size of the model.

Which quantization of Aya 23 35B should I use on the RTX 3090 24GB?

Q2_K. Of the versions that fit, it is the one that runs best; when two run equally well, the check prefers the higher-rated one.

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.