gpt-oss-120b on local hardware
Estimated comfort for repository coding with Aider on a MacBook Pro M4 Pro 48GB.
gpt-oss-120bOpen source
openai · apache-2.0 · 3,181,265 downloads
Mixture of expertsReasoning / thinking mode
- Parameters
- 117B exact
- Active per token
- ≈5.2B (4 of 128 experts)
- Context window
- 4K tokensRoPE-extended to 128K (quality degrades past native — treat long-context ratings as optimistic)
- Attention
- Sliding-window + global layers
- Layers · KV heads · head dim
- 36 · 8 · 64
- KV cache
- 36 KB/token · 1.1 GB at 32K
- Tool calling
- good (from chat template)
- Thinking
- good (from chat template)
- Vision
- basic (inferred — no supporting signal)
- Commercial use
- allowedRetain the required license and attribution notices
- Import confidence
- medium · capabilities estimated
Weights by quantization
MXFP4
59 GB
gguf file size
Q3_K_M
58 GB
gguf file size
Q4_K_M
58 GB
gguf file size
Q5_K_M
59 GB
gguf file size
Q6_K
59 GB
gguf file size
- • Context native 4096, rope-extended to 131072 (yarn); quality degrades past native — treat long-context ratings as optimistic.
- • vision: basic, inferred from missing signals (low confidence).
- • Capability ratings are estimated from model size and name — this model hasn't been assessed by hand.
- • GGUF sizes come from community conversions matched by model name; equivalence and runtime support are not verified. MLX conversion availability is unverified.
gpt-oss-120b — 3-bit
Does not fit in memory
Needs ≈49 GB but only ≈37 GB is available after the OS, other apps and Aider.
High confidence