How to Inspect Local LLM Specs with Ollama’s ‘show’ Command

How to Use

Run ollama show <model_name> in a terminal, replacing <model_name> with the actual model identifier, e.g., phi4-mini:3.8b.

Understanding the Output

The command prints several sections:

Model

architecture

: base architecture type, e.g., phi3. parameters: number of parameters, e.g., 3.8B (38 billion). context length: maximum token context, e.g., 131072. embedding length: dimension of word‑embedding vectors, e.g., 3072. quantization: quantization level, e.g., Q4_K_M, which reduces model size and speeds inference.

Capabilities

completion

: basic text generation/completion. tools: support for external tool or function calling.

License

Shows licensing information, e.g., Microsoft copyright.

Example

Running ollama show phi4-mini:3.8b yields output similar to:

Model
    architecture        phi3
    parameters          3.8B
    context length      131072
    embedding length    3072
    quantization        Q4_K_M

 Capabilities
    completion
    tools

 License
    Microsoft.
    Copyright (c) Microsoft Corporation.

From this we can infer that the model uses the phi3 architecture, has 3.8 billion parameters, supports a context of 131 k tokens, is quantized at Q4_K_M, can generate text and call tools, and is licensed by Microsoft.

Conclusion

Using ollama show lets you quickly retrieve key characteristics of locally stored Ollama models, helping you choose the right model for a task, assess resource requirements, and understand its capabilities.