FitMyLLM
▸ AMD· RDNA 4.0

AMD Radeon RX 9070 XT

Running LLMs on the Radeon RX 9070 XT — the long read: which models fit at which quantisation, and the settings worth changing. · Or what else $599 buys

VRAM
16 GB
MID-RANGE
BANDWIDTH
645
GB/S
MODELS Q4
262/449
58%
7B Q4 SPEED
~82
BLAZING
▸ MODEL COVERAGE @ Q458% OF ALL
▸ ESTIMATED SPEED· BY MODEL SIZE @ Q4

Average speeds at Q4 quantization. Actual performance varies by model architecture and context length.

3B
~191
TOK/S
7B
~82
TOK/S
14B
~41
TOK/S
32B
—
18.0GB NEEDED
70B
—
39.4GB NEEDED
▸ MEASURED RIG REPORTS

We rent the machine and time every model on it: decode, VRAM peak, concurrency, watts and cost per million tokens. Yours may already be one of them — and two are free to read in full.

SEE THE REPORTS →
▸ SPECIFICATIONS
VRAM
16 GB
BANDWIDTH
645 GB/s
FP16 COMPUTE
97.3 TFLOPS
TDP
304W
MEMORY
GDDR6
ARCHITECTURE
RDNA 4.0
STREAM PROCESSORS
4,096
COMPUTE UNITS
64
PCIE
Gen 5 x16
MSRP
$599
260
FAST MODELS · >30 TOK/S
Real-time chat speed
262
USABLE · >10 TOK/S
Comfortable for all tasks
262
TOTAL COMPATIBLE
Fit in VRAM at Q4
▸ WHERE TO BUY

MSRP $599

Some links are affiliate links — we may earn a small commission at no extra cost to you. This helps keep FitMyLLM free and independent.

▸ DON’T WANT TO BUY?

Test Radeon RX 9070 XT (or anything bigger) without committing. Pay by the second, cancel anytime.

Spin up in ~60s. Pay by the second. Cancel anytime.

Some links are affiliate links — we may earn a small commission at no extra cost to you. This helps keep FitMyLLM free and independent.

▸ COMPATIBLE MODELS· 262
S
Falcon-H1R Tiny 90M0.09B
FALCON·256K CTX· CHAT· REASONING
6370
TOK/S · 3% VRAM
›
S
SmolLM2 135M0.135B
SMOLLM·2K CTX· CHAT
4247
TOK/S · 4% VRAM
›
S
nomic-embed-text-v1.5 100M0.14B
EMBEDDING·8K CTX· CHAT
4095
TOK/S · 4% VRAM
›
S
GPT-2 124M0.14B
GPT2·1K CTX· CHAT
4095
TOK/S · 4% VRAM
›
S
SmolVLM 256M0.256B
SMOLLM·8K CTX· CHAT· VISION
2240
TOK/S · 4% VRAM
›
S
Gemma 3 270M0.27B
GEMMA·32K CTX· CHAT
2123
TOK/S · 4% VRAM
›
S
Snowflake Arctic Embed M v2.00.305B
EMBEDDING·8K CTX· EMBEDDING· MULTILINGUAL
1880
TOK/S · 4% VRAM
›
S
bge-large-en-v1.5 335M0.335B
EMBEDDING·1K CTX· CHAT
1711
TOK/S · 4% VRAM
›
S
mxbai-embed-large-v10.335B
EMBEDDING·1K CTX· EMBEDDING
1711
TOK/S · 4% VRAM
›
S
Snowflake Arctic Embed L0.335B
EMBEDDING·1K CTX· EMBEDDING
1711
TOK/S · 4% VRAM
›
S
LFM2 350M0.35B
LFM·125K CTX· CHAT· TOOL_USE
1638
TOK/S · 4% VRAM
›
S
SmolLM2 360M0.36B
SMOLLM·8K CTX· CHAT
1593
TOK/S · 4% VRAM
›
S
GPT-2 Medium 345M0.38B
GPT2·1K CTX· CHAT
1509
TOK/S · 5% VRAM
›
S
Qwen 2.5 0.5B0.5B
QWEN·32K CTX· CHAT
1147
TOK/S · 5% VRAM
›
S
SmolVLM 500M0.5B
SMOLLM·8K CTX· CHAT· VISION
1147
TOK/S · 5% VRAM
›
S
Falcon-H1 0.5B0.52B
FALCON·128K CTX· CHAT
1103
TOK/S · 5% VRAM
›
S
BGE-M30.568B
EMBEDDING·8K CTX· EMBEDDING
1009
TOK/S · 5% VRAM
›
S
Snowflake Arctic Embed L v2.00.568B
EMBEDDING·8K CTX· EMBEDDING· MULTILINGUAL
1009
TOK/S · 5% VRAM
›
S
Qwen3-Embedding 0.6B0.6B
EMBEDDING·32K CTX· EMBEDDING· MULTILINGUAL
956
TOK/S · 5% VRAM
›
S
Falcon-H1R Tiny 0.6B0.6B
FALCON·32K CTX· CHAT· REASONING
956
TOK/S · 5% VRAM
›
S
Falcon Perception 0.6B0.6B
FALCON·4K CTX· VISION
956
TOK/S · 5% VRAM
›
S
Qwen 1.5 0.5B0.62B
QWEN·32K CTX· CHAT
925
TOK/S · 5% VRAM
›
S
LFM2 700M0.74B
LFM·125K CTX· CHAT· TOOL_USE
775
TOK/S · 6% VRAM
›
S
Qwen3 0.6B0.75B
QWEN·32K CTX· CHAT· REASONING
764
TOK/S · 6% VRAM
›
S
GPT-2 Large 774M0.81B
GPT2·1K CTX· CHAT
708
TOK/S · 6% VRAM
›
S
Qwen 3.5 0.8B0.87B
QWEN·256K CTX· CHAT· CODING· MULTILINGUAL
659
TOK/S · 6% VRAM
›
S
InternVL3 1B0.94B
OTHER·32K CTX· CHAT· VISION
610
TOK/S · 7% VRAM
›
S
MiniCPM5 1B1.08B
MINICPM·128K CTX· CHAT· REASONING· MULTILINGUAL
531
TOK/S · 7% VRAM
›
S
TinyLlama 1.1B1.1B
LLAMA·2K CTX· CHAT
521
TOK/S · 7% VRAM
›
S
LFM2.5-1.2B-Thinking1.2B
LFM·122K CTX· CHAT· REASONING· TOOL_USE
478
TOK/S · 8% VRAM
›
S
Llama-3.2-1B1.2B
LLAMA·4K CTX· CHAT
478
TOK/S · 8% VRAM
›
S
LFM2 1.2B1.2B
LFM·125K CTX· CHAT· TOOL_USE· MULTILINGUAL
478
TOK/S · 8% VRAM
›
S
Zamba2 1.2B1.2B
OTHER·4K CTX· CHAT
478
TOK/S · 8% VRAM
›
S
EXAONE-4.0-1.2B1.3B
EXAONE·64K CTX· CHAT
441
TOK/S · 8% VRAM
›
S
OPT 1.3B1.3B
OPT·2K CTX· CHAT
441
TOK/S · 8% VRAM
›
S
MiniCPM-V 4.61.3B
OTHER·256K CTX· CHAT· VISION
441
TOK/S · 8% VRAM
›
S
DeepSeek Coder 1.3B1.35B
DEEPSEEK·16K CTX· CODING
425
TOK/S · 8% VRAM
›
S
Phi-1 1.3B1.42B
PHI·2K CTX· CODING
404
TOK/S · 8% VRAM
›
S
Phi-1.5 1.3B1.42B
PHI·2K CTX· CHAT· CODING
404
TOK/S · 8% VRAM
›
S
LFM2 8B A1B8.3BMoE
LFM·125K CTX· CHAT· TOOL_USE· MULTILINGUAL
382
TOK/S · 35% VRAM
›
▸ NEXT STEP

Get personalized recommendations.

See ranked models with benchmark scores, run commands, and precise speed estimates for your Radeon RX 9070 XT.

WHAT THIS CARD IS WORTH

Radeon RX 9070 XT holds 262 of the models in our catalogue and is, in practice, a IQ4_XS card — the largest it takes is ERNIE 4.5 21B A3B at IQ4_XS.

TOKENS/SEC PER $100
12.0
8B at Q4_K_M, so cards compare like for like
VRAM PER $100
2.67 GB
what memory costs on this card
THE BIGGEST IT TAKES
ERNIE 4.5 21B A3B21.95B · IQ4_XS13.9 GB151 tok/sEST
GPT-OSS 20B21B · Q4_K_S13.8 GB128 tok/sEST
Reka Flash 321B · Q4_K_S14.2 GB32 tok/sEST
Reka Flash 3.121B · Q4_K_S14.2 GB32 tok/sEST
InternLM2 20B19.8B · Q4_K_M14.3 GB32 tok/sEST
InternLM2.5 20B19.8B · Q4_K_M14.3 GB32 tok/sEST
Ling-lite 16.8B16.8B · Q5_K_M13.6 GB148 tok/sEST
DeepSeek V2 Lite 16B16B · Q5_K_M12.9 GB148 tok/sEST
AGAINST CARDS OF SIMILAR MEMORY
M1 Max (32GB)21 GB$1,4993.0 tok/s per $100
M2 Max (32GB)21 GB$1,7992.5 tok/s per $100
M2 Pro (32GB)21 GB$1,4991.9 tok/s per $100
M4 (32GB)21 GB$1,1991.7 tok/s per $100

Neighbours in memory rather than in price: memory decides whether a card can do the job at all, so two cards of the same size at different prices is the comparison you are making. Ordered by memory, then bandwidth — not by the value column, which is worked out from bandwidth and price alone and therefore rewards a cheap card whatever its software stack does to that bandwidth in practice. Read it as one input, not as a ranking.

▸ DEVICE UNDER TEST

AMD Radeon RX 9070 XT — 16 GB VRAM.

▸ RADEON RX 9070 XT SPEC
BRAND
AMD
VRAM
16 GB GDDR6
BANDWIDTH
645 GB/s
FP16 COMPUTE
97.3 TFLOPS
FP32 COMPUTE
48.7 TFLOPS
STREAM PROCESSORS
4,096
TDP
304 W
ARCHITECTURE
RDNA 4.0
MSRP
$599
▸ AI CAPABILITY
262/ 449 models @ Q4

With 16 GB VRAM and 645 GB/s bandwidth, this GPU handles models up to 19.8B parameters.

Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~74 tok/s.

RENT IT — LIVE PRICES
Checking what the clouds are charging…
§ 01TOP MODELS FOR RADEON RX 9070 XT
262 FIT · SHOWING 20
MODELSIZEVRAM Q4TOK/SAVG
InternLM2 20B19.8B12.6 GB2945.1
InternLM2.5 20B19.8B12.6 GB2950.9
Ling-lite 16.8B16.8B10.8 GB239—
DeepSeek V2 Lite 16B16B10.3 GB23938.0
StarCoder2 15B15.96B10.2 GB3626.5
DeepSeek-Coder-V2-Lite 15.7B15.7B10.1 GB23943.0
DeepSeek-VL2 Small 16B15.7B10.1 GB23943.1
StarCoder 15B15.5B10.0 GB3721.0
InternVL3 14B15.12B9.7 GB3838.1
Phi-4-reasoning-vision 15B15B9.7 GB3842.8
DeepSeek R1 Distill Qwen 14B14.8B9.5 GB3943.9
DeepCoder 14B14.8B9.5 GB3938.7
Qwen2.5-Coder-14B14.8B9.5 GB3941.3
Qwen2.5-14B14.8B9.5 GB3941.3
Qwen3 14B14.8B9.5 GB3945.7
phi-4 14B14.66B9.4 GB3933.7
Phi-4-reasoning 14B14.66B9.4 GB3933.7
Phi-4-reasoning-plus 14B14.66B9.4 GB3975.5
Ministral 3 14B14B9.0 GB4125.9
Phi-3-medium-14b14B9.0 GB4133.7