Describe the feature
I hope to add a community benchmark sharing feature:
Users can submit their real‑world test data, including:
- Machine hardware config: GPU model, VRAM, CPU, RAM
- Model name & quant version running on FreeToken
- Real measured performance: token/s prefill/decode speed
- Max context length achieved, stable or limitation
These submitted benchmark records can be browsed/searchable for other users.
Use case
Users who are making hardware purchase decisions can directly look up real‑world test results from other community users.
Help people know what performance/context size to expect on their hardware before buying GPU or running large models.
Alternative ideas considered
Option A: Maintain a markdown table in repo docs for manual PR contributions.
Option B: Simple UI inside desktop app to upload benchmark entries.
Option C: Discussion page to collect benchmark reports first.
Describe the feature
I hope to add a community benchmark sharing feature:
Users can submit their real‑world test data, including:
These submitted benchmark records can be browsed/searchable for other users.
Use case
Users who are making hardware purchase decisions can directly look up real‑world test results from other community users.
Help people know what performance/context size to expect on their hardware before buying GPU or running large models.
Alternative ideas considered
Option A: Maintain a markdown table in repo docs for manual PR contributions.
Option B: Simple UI inside desktop app to upload benchmark entries.
Option C: Discussion page to collect benchmark reports first.