AMD · RDNA 3

Best AI Models for AMD Radeon RX 7900 XT (20 GB)

Runs 27–31B models well; 35B MoE only just fits.

20 GB GDDR6800 GB/s103 TFLOPS FP16315 WRDNA 3 2022

What LLMs can the RX 7900 XT run?

#	Model	Speed	Quant · VRAM
1	Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 27.6 t/s	27.6 tok/s	Q4_K_M · 17.9 GB
2	Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 27.6 t/s	27.6 tok/s	Q4_K_M · 17.9 GB
3	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 85.3 t/s	85.3 tok/s	Q4_K_M · 16.5 GB
4	GPT OSS 20BOpenAI Q4_K_M · 13.1 GB · 98.1 t/s	98.1 tok/s	Q4_K_M · 13.1 GB
5	LFM2.5 24B A2BLiquid AI Q4_K_M · 15.0 GB · 89.5 t/s	89.5 tok/s	Q4_K_M · 15.0 GB

#	Model	Speed	Quant · VRAM
1	Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 27.6 t/s	27.6 tok/s	Q4_K_M · 17.9 GB
2	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 85.3 t/s	85.3 tok/s	Q4_K_M · 16.5 GB
3	Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 27.6 t/s	27.6 tok/s	Q4_K_M · 17.9 GB
4	GPT OSS 20BOpenAI Q4_K_M · 13.1 GB · 98.1 t/s	98.1 tok/s	Q4_K_M · 13.1 GB
5	LFM2.5 24B A2BLiquid AI Q4_K_M · 15.0 GB · 89.5 t/s	89.5 tok/s	Q4_K_M · 15.0 GB

#	Model	Speed	Quant · VRAM
1	Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 27.6 t/s	27.6 tok/s	Q4_K_M · 17.9 GB
2	Qwen3.5 27BAlibaba Q4_K_M · 17.9 GB · 27.6 t/s	27.6 tok/s	Q4_K_M · 17.9 GB
3	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 85.3 t/s	85.3 tok/s	Q4_K_M · 16.5 GB
4	GPT OSS 20BOpenAI Q4_K_M · 13.1 GB · 98.1 t/s	98.1 tok/s	Q4_K_M · 13.1 GB
5	LFM2.5 24B A2BLiquid AI Q4_K_M · 15.0 GB · 89.5 t/s	89.5 tok/s	Q4_K_M · 15.0 GB

#	Model	Speed	Math Score	Quant · VRAM
1	Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 27.6 t/s	27.6 tok/s		Q4_K_M · 17.9 GB
2	Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 27.6 t/s	27.6 tok/s		Q4_K_M · 17.9 GB
3	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 85.3 t/s	85.3 tok/s	overall	Q4_K_M · 16.5 GB
4	LFM2.5 24B A2BLiquid AI Q4_K_M · 15.0 GB · 89.5 t/s	89.5 tok/s		Q4_K_M · 15.0 GB
5	MiMo V2.6 Distill Qwen 9BXiaomi Q4_K_M · 6.5 GB · 77.3 t/s	77.3 tok/s		Q4_K_M · 6.5 GB

#	Model	Speed	Quant · VRAM
1	Qwen3.5 9BAlibabaHow to run → Q4_K_M · 6.6 GB · 75.5 t/s	75.5 tok/s	Q4_K_M · 6.6 GB
2	Qwen3.5 4BAlibaba Q4_K_M · 3.4 GB · 147 t/s	147 tok/s	Q4_K_M · 3.4 GB
3	GPT OSS 20BOpenAI Q4_K_M · 13.1 GB · 98.1 t/s	98.1 tok/s	Q4_K_M · 13.1 GB
4	Mistral Nemo Instruct 2407Mistral AI Q4_K_M · 11.0 GB · 59.5 t/s	59.5 tok/s	Q4_K_M · 11.0 GB
5	LFM2.5 24B A2BLiquid AI Q4_K_M · 15.0 GB · 89.5 t/s	89.5 tok/s	Q4_K_M · 15.0 GB

#	Model	Speed ↓	Quant · VRAM
1	Qwen3.5 4BAlibabaHow to run → Q4_K_M · 3.4 GB · 147 t/s	147 tok/s	Q4_K_M · 3.4 GB
2	GPT OSS 20BOpenAI Q4_K_M · 13.1 GB · 98.1 t/s	98.1 tok/s	Q4_K_M · 13.1 GB
3	LFM2.5 24B A2BLiquid AI Q4_K_M · 15.0 GB · 89.5 t/s	89.5 tok/s	Q4_K_M · 15.0 GB
4	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 85.3 t/s	85.3 tok/s	Q4_K_M · 16.5 GB
5	Qwen3.5 9BAlibaba Q4_K_M · 6.6 GB · 75.5 t/s	75.5 tok/s	Q4_K_M · 6.6 GB

#	Model	Speed	Quant · VRAM
1	Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 27.6 t/s	27.6 tok/s	Q4_K_M · 17.9 GB
2	Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 27.6 t/s	27.6 tok/s	Q4_K_M · 17.9 GB
3	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 85.3 t/s	85.3 tok/s	Q4_K_M · 16.5 GB
4	Gemma 3 12B ITGoogle Q4_K_M · 10.9 GB · 59.7 t/s	59.7 tok/s	Q4_K_M · 10.9 GB
5	MiMo V2.6 Distill Qwen 9BXiaomi Q4_K_M · 6.5 GB · 77.3 t/s	77.3 tok/s	Q4_K_M · 6.5 GB

See all 318 models on the RX 7900 XT →Best quant that fits · speeds estimated on this card

About the RX 7900 XT

The RX 7900 XT is a solid mid-range card in the 16–20 GB class: it runs 27–31B models well. It is more efficient per watt than most 16–20 GB cards, and at its $899 MSRP it gave the best speed per dollar of the cards compared. RDNA 3 generation, released 2022.

Speed in 16–20 GB#3800 GB/s

Per watt#60.24 tok/s per W on Qwen3.5 9B (9B params)

Per dollar#18.4 tok/s per $100 on Qwen3.5 9B (9B params)

GenerationRDNA 3released 2022

RX 7900 XT strengths and weaknesses

Strengths

Runs 35B MoE models: Qwen3.6 35B A3B at 74.5 tok/s in 19.8 GB

Weaknesses

70B models don't fit: Llama 3.3 70B Instruct needs 51.3 GB at Q4 vs 20 GB here
Slow prompt reading: A 4K-token prompt on Qwen3.8 27B waits 10.0 s for the first token
ROCm, not CUDA: No FP8 weights; MXFP4 only through llama.cpp, not vLLM

Speed per dollar and per watt

NewUsed

On Qwen3.5 9B (9B params) at Q4_K_M. New prices are compared with new, used with used. The last two columns are the speed you get per $100 of price and per 100 W of board power; higher is better. The RX 7900 XT has no current price, so it is shown at its MSRP.

Card	New price	tok/s	tok/s per $100	tok/s per 100 W
RX 7900 XT20 GB	about $899	75.5	8.4	24.0
RTX 507012 GB	about $840	68.7	8.2	27.5
RX 7900 XTX24 GB	about $1,130	90.6	8.0	25.5
RTX 5070 Ti16 GBBest new value	about $1,170	91.6	7.8	30.5
RX 907016 GB	about $800	60.4	7.6	27.5
RX 9070 XT16 GB	about $815	60.4	7.4	19.9
RTX 508016 GB	about $1,690	98.1	5.8	27.3
RTX 509032 GB	about $5,376	183	3.4	31.8

Card	Used price	tok/s	tok/s per $100	tok/s per 100 W
RTX 3060 12GB12 GBBest used value	about $329	36.8	11.2	21.6
RX 7900 XTX24 GB	about $1,000	90.6	9.1	25.5
RTX 5070 Ti16 GB	about $1,100	91.6	8.3	30.5
RTX 4070 Ti SUPER16 GB	about $900	68.7	7.6	24.1
RTX 408016 GB	about $1,100	73.3	6.7	22.9
RTX 4080 SUPER16 GB	about $1,155	75.2	6.5	23.5
RTX 309024 GB	about $1,500	95.7	6.4	27.3
RTX 509032 GB	about $6,400	183	2.9	31.8

Similar GPUs

Cards people weigh against the RX 7900 XT, each with the one reason it is worth a look. Speeds on Qwen3.5 9B.

4 GB more and 20% fasterRX 7900 XTXAMD · 24 GB · RDNA 3

▲ 20% speed▲ 4 GB$1,130

12 GB more and 2.4× the speedRTX 5090NVIDIA · 32 GB · Blackwell

▲ 143% speed▲ 12 GB$5,376rent $0.50/hr

4 GB more and 27% fasterRTX 3090NVIDIA · 24 GB · Ampere

▲ 27% speed▲ 4 GB$1,499rent $0.14/hr

As fast, 16 GBRTX 4080 SUPERNVIDIA · 16 GB · Ada Lovelace

same speed▼ 4 GB$999rent $0.23/hr

4 GB more for bigger modelsTITAN RTXNVIDIA · 24 GB · Turing

▼ 9% speed▲ 4 GB$2,499rent $0.22/hr

9% slower, 16 GBRTX 4070 Ti SUPERNVIDIA · 16 GB · Ada Lovelace

▼ 9% speed▼ 4 GB$799rent $0.15/hr

RX 7900 XT tokens per second by model size

Output speed depends on memory bandwidth (800 GB/s). Grey rows need more than 20 GB.

9BQwen3.5 9B · Q4_K_M · 6.6 GB
75.5tok/s
14BQwen3 14B · Q4_K_M · 11.9 GB
50.5tok/s
27–31BQwen3.8 27B · Q4_K_M · 17.9 GB
27.6tok/s
35B MoEQwen3.6 35B A3B · IQ4_XS · 19.8 GB tight · IQ4_XS
74.5tok/s
70BLlama 3.3 70B Instruct · not at 4-bit
needs 51.3 GB
120B MoEGPT OSS 120B · not at 4-bit
needs 70.8 GB

Estimated speeds: Q4_K_M, 4K context, one user. How we estimate

All 1117 models on the RX 7900 XT

Featured only

Best quant that fits, ordered by llmrun Score. Fit grades how well it runs here, S best. How ranking works

LLM models compatible with AMD Radeon RX 7900 XT — ranked by Score, recency and Fit
Model	Quant	VRAM	Speed	Context	Fit
OmniSVG1.1 8B8BChat BF16·29.2 t/s	BF16	17.2 GB86%	29.2 t/s	128K	B62, Decent
Llama Poro 2 8B Instruct8.0BChat Q4_K_M·89.1 t/s	Q4_K_M	6.2 GB31%	89.1 t/s	8K	S89, Runs great
C4ai Command R7b 12 20248.0BChatFunctions Q4_K_M·90.6 t/s	Q4_K_M	7.9 GB39%	90.6 t/s	—	S90, Runs great
Gemma 2 9B IT9.2BChat Q4_K_M·78.7 t/s	Q4_K_M	7.3 GB36%	78.7 t/s	8K	S87, Runs great
Tiny LLM13MChat Q4_K_M·1548.4 t/s	Q4_K_M	0.3 GB2%	1548.4 t/s	1K	S100, Runs great
Sarvam 30B32.2BChatFunctions Q4_K_M·104.7 t/s	Q4_K_M	19.9 GB100%	104.7 t/s	131K	C44, Usable
Param 1 2.9B Instruct2.9BChat BF16·76.3 t/s	BF16	7.1 GB36%	76.3 t/s	8K	S87, Runs great
EuroLLM 1.7B Instruct1.7BChat Q4_K_M·320.0 t/s	Q4_K_M	1.7 GB9%	320.0 t/s	4K	S100, Runs great
Qwen2.5 1.5B Instruct1.5BChatFunctions Q4_K_M·375.0 t/s	Q4_K_M	1.7 GB8%	375.0 t/s	33K	S100, Runs great
LLaMmlein 1B Prerelease1.1BChat Q4_K_M·475.2 t/s	Q4_K_M	1.0 GB5%	475.2 t/s	2K	S100, Runs great
Llama 3.2 1B Instruct1.2BChatFunctions Q4_K_M·585.4 t/s	Q4_K_M	2.3 GB12%	585.4 t/s	131K	S100, Runs great
Qwen2.5 1.5B Quantized.w8a81.8BChatFunctions Q4_K_M·335.7 t/s	Q4_K_M	1.8 GB9%	335.7 t/s	33K	S100, Runs great
ERNIE 4.5 0.3B Paddle361MChat BF16·461.5 t/s	BF16	1.2 GB6%	461.5 t/s	131K	S100, Runs great
Falcon Mamba Tiny Dev9MChat BF16·24000.0 t/s	BF16	0.4 GB2%	24000.0 t/s	—	S100, Runs great
Typhoon Ocr1.5 2B2.1BVisionFunctions Q4_K_M·265.2 t/s	Q4_K_M	3.5 GB17%	265.2 t/s	262K	S100, Runs great

RX 7900 XT specs, power and PSU

Specs

Architecture: RDNA 3
VRAM: 20 GB GDDR6
Memory bandwidth: 800.0 GB/s
FP16 (dense): 103.20 TFLOPS
Stream processors: 5,376
TDP: 315 W
Released: 2022-12-13
MSRP: $899

RX 7900 XT FAQ

Is the RX 7900 XT still worth it in 2026?

At its MSRP of $899 you get 8.4 tok/s per $100 on Qwen3.5 9B, which ranks 1st of the 8 cards compared here; it is also the 3rd fastest out of 18 cards with 16–20 GB. It runs 27–31B models well; 35B MoE only just fits.

What is the largest model the RX 7900 XT can run?

Qwen3.6 35B A3B (36.0B) is the largest at 4-bit or better: at IQ4_XS it uses 19.8 GB of 20 GB, and runs at about 74.5 tok/s.

Can the RX 7900 XT run Qwen3.8 27B?

Yes. Qwen3.8 27B fits at Q4_K_M, using 17.9 GB of 20 GB, and it is the best model that runs well on this card.

Can the RX 7900 XT run 70B models?

No. Llama 3.3 70B Instruct needs 51.3 GB at Q4, more than the 20 GB on this card.

How many tokens per second does the RX 7900 XT give?

About 27.6 tok/s on Qwen3.8 27B at Q4_K_M, with the first token after 10.0 s on a 4K-token prompt.

What power supply does the RX 7900 XT need?

Plan a 650 W PSU (2× the 315 W TDP).

Is 20 GB enough for coding and agents?

Yes for coding: Qwen3.8 27B, the best of the 6 coding models that fit, runs at about 28 tok/s. For agents, not yet: no agent-scored model fits in less than 96 GB.

Is renting cheaper than buying an RX 7900 XT?

An RTX 3090 (24 GB) rents from $0.14/hr. Renting costs as much as buying after ≈6,300 hours (≈4.3 years at 4 h a day).

Best coding LLMs 16 GB VRAM 24 GB VRAM GPUs with 20 GB or more RTX 3090 Ti RTX 5090 Laptop GPU