Llama-3.2-1B-Instruct on local hardware
Estimated comfort for repository coding with Aider on a MacBook Pro M4 Pro 48GB.
Llama-3.2-1B-InstructOpen weights (restricted)
meta-llama · llama3.2 · 5,072,769 downloads
Gated
- Parameters
- 1.2B exact
- Context window
- 8K tokensRoPE-extended to 128K (quality degrades past native — treat long-context ratings as optimistic)
- Attention
- Standard (full attention)
- Layers · KV heads · head dim
- 16 · 8 · 64
- KV cache
- 32 KB/token · 1.0 GB at 32K
- Tool calling
- good (from chat template)
- Thinking
- basic (from chat template)
- Vision
- basic (inferred — no supporting signal)
- Commercial use
- conditionalCommunity license and acceptable-use restrictions apply; large services need separate permission
- Import confidence
- medium · capabilities estimated
Weights by quantization
Q3_K_M
0.7 GB
gguf file size
Q4_K_M
0.8 GB
gguf file size
Q5_K_M
0.8 GB
gguf file size
Q6_K
1.0 GB
gguf file size
Q8_0
1.2 GB
gguf file size
F16 / BF16
2.3 GB
gguf file size
- • Architecture read from unsloth/Llama-3.2-1B-Instruct.
- • Context native 8192, rope-extended to 131072 (llama3); 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.
- • Commercial use conditional: Community license and acceptable-use restrictions apply; large services need separate permission.
- • GGUF sizes come from community conversions matched by model name; equivalence and runtime support are not verified. MLX conversion availability is unverified.
Llama-3.2-1B-Instruct — 6-bit
Technically Runs for your workload
Speed
130–180 tok/s
Excellent
Memory left
35 GB
of 48 GB
First word in
≈4 s
Excellent
All the details
- Your setup
- MacBook Pro M4 Pro 48 GB
- Workload
- Repository coding with Aider
- Runtime → tool
- LM Studio → Aider
- Memory used
- 2.0 GB
- Context
- 8K · up to 8K practical
- Tool support
- Official
- Use-case fit
- Poorly suited
Why?
- Model fits with approximately 35 GB of memory headroom.
- LM Studio supports hardware acceleration on this machine (Metal (Apple GPU)).
- Aider can connect to LM Studio (officially supported).
- Generation (130–180 tok/s) should be appropriate for interactive coding.
- Ingesting a prompt of 7K tokens takes ≈4 s; prompt processing keeps up with interactive coding.
- 8K window; generation drops to 130–180 tok/s when it is full. The model is limited to 8K.
- Llama-3.2-1B-Instruct supports at most 8K context; 32K was requested.
- Weak coding capability for interactive coding.
Why “Technically Runs” rather than “Borderline”?
- Workload suitability: Weak coding capability for interactive coding.
- Context: 8K window; generation drops to 130–180 tok/s when it is full. The model is limited to 8K.
How sure are we?
Medium confidenceEstimated from bandwidth and compute, calibrated against 2 verified benchmarks of other models on this same chip (measured with a different engine, so calibration is partial).
- Calibrated with verified benchmarks of other models on this chip.
Medium confidence