Ling-3.0-flash-UBTree on local hardware
Estimated comfort for repository coding with Aider on a MacBook Pro M4 Pro 48GB.
Ling-3.0-flash-UBTreeOpen source
inclusionAI · mit · 836 downloads
- Parameters
- 0.5B exact
- Context window
- 128K tokens
- Attention
- Standard (full attention)
- Layers · KV heads · head dim
- 5 · 32 · 128
- KV cache
- 80 KB/token · 2.5 GB at 32K
- Tool calling
- basic (inferred — no supporting signal)
- Thinking
- basic (inferred — no supporting signal)
- Vision
- basic (inferred — no supporting signal)
- Commercial use
- allowedRetain the required license and attribution notices
- Import confidence
- medium · capabilities estimated
Weights by quantization
Q3_K_M
0.2 GB
computed · low confidence
Q4_K_M
0.3 GB
computed · low confidence
Q5_K_M
0.3 GB
computed · low confidence
Q6_K
0.4 GB
computed · low confidence
Q8_0
0.5 GB
computed · low confidence
F16 / BF16
0.9 GB
safetensors file size
- • Assumes GGUF / MLX conversions of this model exist (they do for most popular models). Sizes are computed from bits per weight.
- • No chat template found; tool calling is inferred as limited (low confidence).
- • toolCalling: basic, inferred from missing signals (low confidence).
- • thinking: basic, inferred from missing signals (low confidence).
- • 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.
Ling-3.0-flash-UBTree — 8-bit
Technically Runs for your workload
Speed
90–120 tok/s
Excellent
Memory left
33 GB
of 48 GB
First word in
≈3 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
- 3.9 GB
- Context
- 32K · up to 128K practical
- Tool support
- Official
- Use-case fit
- Poorly suited
Why?
- Model fits with approximately 33 GB of memory headroom.
- LM Studio supports hardware acceleration on this machine (Metal (Apple GPU)).
- Aider can connect to LM Studio (officially supported).
- Generation (90–120 tok/s) should be appropriate for interactive coding.
- Ingesting a prompt of 14K tokens takes ≈3 s; prompt processing keeps up with interactive coding.
- 32K working context is reasonable.
- Weak coding capability for interactive coding.
Why “Technically Runs” rather than “Borderline”?
- Workload suitability: Weak coding capability for interactive coding.
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