Skip to content

Can Radeon RX 7900 XTX run Kolibri-1 78B-A3.5B?

Comfortable: GGUF IQ1_M needs 20.9 GB at 4K, batch 1; headroom 3.1 GB.

Text input, FP16 KV cache, one card. Unified-memory capacity uses the GPU guide’s usable limit; negative headroom may require host offload. Speed source settings may differ from this 4K estimate.

Measured runs

No measured run for this pair. Sizing is an estimate, not a runtime test.

Quant memory at 4K

QuantTotal GBVerdictWeight sizing
GGUF IQ1_M20.9ComfortableVerified file
GGUF IQ2_XXS24.0TightVerified file
GGUF IQ2_XS25.6Does not fitVerified file
GGUF IQ3_XXS35.4Does not fitVerified file
GGUF Q3_K_M39.5Does not fitVerified file
GGUF IQ4_XS45.8Does not fitVerified file
GGUF Q4_K_M52.2Does not fitVerified file
GGUF Q8_087.1Does not fitVerified file

Context capacity · GGUF IQ1_M

ContextTotal GBVerdict
4K20.9Comfortable
16K21.2Tight
32K21.5Tight
128K23.6Tight

Generation speed

GGUF IQ1_M: 1138 tok/s · Bandwidth ceiling (not measured)

Other model / GPU pairs