Can I run clef-flash on a Radeon RX 7900 XTX?

Yes, just

Its published weights (18.8 GB) fit in the 21.6 GB a Radeon RX 7900 XTX can use, with room for ≈ 81K tokens of context.

Full analysis of clef-flash Everything that runs on Radeon RX 7900 XTX Get the weights on Hugging Face

hf download Cloudflare/clef-flash

Files and context on Radeon RX 7900 XTX

Each file of clef-flash (and of GGUF versions the site has analyzed), and how much context fits next to it, with the context cache in 16-bit or 8-bit.

File	Size	On Radeon RX 7900 XTX	Context (16-bit)	Context (8-bit)
As published	18.8 GB	Tight	81K	163K
BF16ggml-org/Clef-Flash-GGUF	18.2 GB	Tight	101K	202K
Q8_0ggml-org/Clef-Flash-GGUF	9.66 GB	Fits	256K	256K
Q4_K_Mggml-org/Clef-Flash-GGUF	6.49 GB	Fits	256K	256K
BF16bartowski/Cloudflare_clef-flash-GGUF	18.2 GB	Tight	101K	202K
Q8_0bartowski/Cloudflare_clef-flash-GGUF	9.68 GB	Fits	256K	256K
Q6_K_Lbartowski/Cloudflare_clef-flash-GGUF	8.32 GB	Fits	256K	256K
Q6_Kbartowski/Cloudflare_clef-flash-GGUF	7.83 GB	Fits	256K	256K
Q5_K_Mbartowski/Cloudflare_clef-flash-GGUF	6.98 GB	Fits	256K	256K
Q4_K_Lbartowski/Cloudflare_clef-flash-GGUF	6.80 GB	Fits	256K	256K
Q5_K_Sbartowski/Cloudflare_clef-flash-GGUF	6.66 GB	Fits	256K	256K
Q4_1bartowski/Cloudflare_clef-flash-GGUF	6.07 GB	Fits	256K	256K
Q4_K_Mbartowski/Cloudflare_clef-flash-GGUF	6.04 GB	Fits	256K	256K
Q4_K_Sbartowski/Cloudflare_clef-flash-GGUF	5.73 GB	Fits	256K	256K
IQ4_NLbartowski/Cloudflare_clef-flash-GGUF	5.63 GB	Fits	256K	256K
Q4_0bartowski/Cloudflare_clef-flash-GGUF	5.60 GB	Fits	256K	256K
IQ4_XSbartowski/Cloudflare_clef-flash-GGUF	5.40 GB	Fits	256K	256K
Q3_K_Lbartowski/Cloudflare_clef-flash-GGUF	5.24 GB	Fits	256K	256K
Q3_K_Mbartowski/Cloudflare_clef-flash-GGUF	5.05 GB	Fits	256K	256K
IQ3_Mbartowski/Cloudflare_clef-flash-GGUF	4.85 GB	Fits	256K	256K
Q3_K_Sbartowski/Cloudflare_clef-flash-GGUF	4.80 GB	Fits	256K	256K
Q2_Kbartowski/Cloudflare_clef-flash-GGUF	4.19 GB	Fits	256K	256K

clef-flash on other hardware

Popular models that run on Radeon RX 7900 XTX

Assumes 90% of the card's memory is usable and at least 4K tokens of context for one conversation, with the context cache in 16-bit unless noted. Fits leaves at least 20% of that memory for context; Tight leaves less.