How much VRAM does OTel 2.0 LLM 31B Instruct need on a Radeon RX 9070 XT?

At Q4_K_M with an 8k context it needs 19.92 GB in total: 18.06 GB of weights, 0.94 GB of KV cache and 0.93 GB of overhead. The Radeon RX 9070 XT makes 15.2 GB available once the operating system has taken its share.

Can an RX 9070 XT run OTel 2.0 LLM 31B Instruct?

Q: Can an RX 9070 XT run OTel 2.0 LLM 31B Instruct?

Only barely. It loads at IQ3_XXS, which compresses the weights far enough to visibly degrade the model, so the honest answer is that this pairing does not work.

The numbers

OTel 2.0 LLM 31B Instruct has 32.11 billion parameters across 60 layers. The Radeon RX 9070 XT has 16 GB of VRAM at 645 GB/s, of which about 15.2 GB is left for a model once the operating system has taken its share.

Quantisation	Total needed	Fits	Speed
Q8_0	33.64 GB	no	—
Q6_K	26.39 GB	no	—
Q5_K_M	23.14 GB	no	—
Q4_K_M	19.92 GB	no	—
IQ4_XS	17.75 GB	no	—
Q3_K_M	16.48 GB	no	—
IQ3_XXS degraded	13.31 GB	yes	37.5 t/s
Q2_K degraded	14.39 GB	yes	34.5 t/s
IQ2_XXS degraded	9.57 GB	yes	53.8 t/s

3 of the 9 levels fit. Levels marked degraded are listed for completeness, not as advice.

What to do instead

The straightforward answer is a smaller model from the same family. These do fit on a Radeon RX 9070 XT, at a quantisation worth using:

Agents A1 4B — 4.54B at Q8_0, 6.25 GB, about 84.6 tokens/s
OneRec 1.7B — 2.13B at Q8_0, 3.71 GB, about 155.7 tokens/s

Shortening the context window will not rescue this one: at 4k it still wants 19.76 GB against 19.92 GB at 8k, because the weights rather than the cache are what fill the card. The gap here is too large for settings to close.

Or the hardware that does run it

The cheapest device we track that runs OTel 2.0 LLM 31B Instruct at a quantisation worth using is the Radeon RX 7900 XT, around $650 , running it at IQ4_XS at about 34.2 tokens per second. Check the current price (affiliate link; indicative price reviewed 2026-09-01). Compare every option.

See how fast it feels

Radeon RX 9070 XT running OTel 2.0 LLM 31B Instruct at IQ3_XXS

Wait for the first word1.2 s

Then writes at37.5 tok/s

Whole answer5.8 s

YouWhy does my model use more memory when the conversation gets longer?

Model

Because of the KV cache. Every token you send leaves behind a key and a value vector in each layer of the model, and those stay in memory for as long as the conversation lasts.

The weights are a fixed cost: load a 4-bit 8B model and that is about 4.8 GB, whether you write one word or ten thousand. The cache is the part that grows, and it grows in a straight line with the number of tokens in the window.

How steeply depends on the model's attention design. With grouped-query attention, several query heads share one key-value pair, which cuts the cache by that ratio. Without it, every head keeps its own, and a long context can cost more memory than the weights themselves.

If you are short on memory, the first thing to try is lowering the context window in your runtime, not the quantisation.

Simulated from our estimate at a 512-token question, not a recording. Assumes nothing else is competing for the GPU. The same answer on RTX 4090 at Q5_K_M would take about 4.9 s.

Keep going

Everything the Radeon RX 9070 XT runs, the full breakdown for OTel 2.0 LLM 31B Instruct, or check a different pairing in the calculator.