Can I run it?

Can I run Qwen3 Coder 30B A3B Instruct on GeForce RTX 5060 Ti 8GB?

Expert offload

Expert offload on your GeForce RTX 5060 Ti 8GB: Q4_K_M keeps shared weights on the GPU and about 13.9 GB of experts in system RAM. About 10–14 tok/s. Active-parameter speed is approximate.

Qwen3 Coder 30B A3B Instruct details · GeForce RTX 5060 Ti 8GB page · How this estimate is made

Run commands

# Expert offload is approximate. llama.cpp can pin experts to CPU:
#   -ot "\\.ffn_.*_exps\\.=CPU"
llama-server -hf unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF:Q4_K_M -c 8192 -ngl 99

ollama run qwen3

lms get unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF
# Then load quant Q4_K_M

Assumptions

• Estimate from memory bandwidth, not a benchmark. Decode speed uses 55–75% of peak bandwidth.
• KV cache is FP16. Engine overhead assumes llama.cpp (0.3 GB + 1% of the weights). Ollama is closer to 0.9 GB + 3%, LM Studio closer to 1 GB + 3%.
• Q4_K_M size is the real GGUF file size from Hugging Face.
• 0.5 GB is reserved because this GPU also drives a display, then 5% of the rest is left free.
• Speed is quoted at 8K context, not at the maximum context that still fits.
• Mixture-of-experts split is approximate: shared layers on the GPU, expert tensors in system RAM. Speed reads more than the active-parameter slice, because shared layers stay on the bus.

Models that fit this machine more comfortably

• Devstral Small 2 24B Instruct 2512 — Heavily degraded, UD-IQ1_M, 37–50 tok/s (estimate)
• Qwen2.5 Coder 14B Instruct 128K — Tight fit, Q2_K, 37–51 tok/s (estimate)
• Qwen2.5 Coder 7B Instruct 128K — Tight fit, Q6_K, 37–50 tok/s (estimate)

Updated 2026-10-24. Profile on this page is the device default plus 32 GB of DDR5-5600 system RAM.