RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can gpt oss 20b run on the Mac M3 24GB?

Yes. At 8K context the best fit is F16, using about 13.8 GB of the 16.1 GB usable on the Mac M3 24GB. Estimated speed ceiling: ~36 tokens/sec.

Memory by quant and context

Quant4K ctx8K ctx32K ctx128K ctxMax tok/s (est., 8K)
2-bit (Q2_K) •11.6 GB Fits11.7 GB Fits12.2 GB Fits14.5 GB Fits43
3-bit (Q3_K_M) •11.6 GB Fits11.7 GB Fits12.3 GB Fits14.5 GB Tight43
4-bit (Q4_K_M) •11.7 GB Fits11.8 GB Fits12.4 GB Fits14.6 GB Tight43
5-bit (Q5_K_M) •11.8 GB Fits11.9 GB Fits12.4 GB Fits14.7 GB Tight43
6-bit (Q6_K) •12.1 GB Fits12.2 GB Fits12.8 GB Fits15.0 GB Tight41
8-bit (Q8_0) •12.2 GB Fits12.3 GB Fits12.8 GB Fits15.1 GB Tight41
16-bit (F16/BF16) •13.7 GB Fits13.8 GB Fits14.4 GB Fits16.6 GB No36

• real GGUF file size from unsloth/gpt-oss-20b-GGUF. Speed is a bandwidth ceiling (100 GB/s), not a benchmark. Methodology.

More

gpt oss 20b on every GPU and Mac · Every model that runs on the Mac M3 24GB · Apple Silicon guide · Try other settings in the calculator

Data: Hugging Face Hub (openai/gpt-oss-20b), Apple specs. Model data last checked 2026-11-17; refreshed hourly.