Hi everyone, I made whatmodel. It's a simple site that can be visited in the browser or used from the terminal via a:
curl whatmodel.sh
To use it,
1. Enter your RAM and GPU
2. See what models you can run locally with its rank and the command to run it
3. Compare them with API models on the same score
Building it made me realise a 32GB Mac can now run a model that scores on par with GPT-5, completely locally and for free (e.g. Qwen 3.8 27B at 4-bit).
Each model gets one score fitted across 18,731 published benchmark results, and every result links to its source.
The method is at https://whatmodel.sh/method and the data, scoring code and bench script are open source: https://github.com/jakemanger/whatmodel
If you'd like to contribute, there's a bench script to collect real-world tok/s and you can also submit any new models and benchmark scores along with their source on github.
curl whatmodel.sh
To use it,
1. Enter your RAM and GPU
2. See what models you can run locally with its rank and the command to run it
3. Compare them with API models on the same score
Building it made me realise a 32GB Mac can now run a model that scores on par with GPT-5, completely locally and for free (e.g. Qwen 3.8 27B at 4-bit).
Each model gets one score fitted across 18,731 published benchmark results, and every result links to its source.
The method is at https://whatmodel.sh/method and the data, scoring code and bench script are open source: https://github.com/jakemanger/whatmodel
If you'd like to contribute, there's a bench script to collect real-world tok/s and you can also submit any new models and benchmark scores along with their source on github.