# Measured Local AI Fit Receipts

- canonical_url: https://bmdpat.com/receipts
- markdown_url: https://bmdpat.com/receipts.md
- last_updated: 2026-08-08
- answer_kind: measured
- source: owned hardware benchmark artifacts listed on each receipt

These receipts are measured data points, not guarantees for every runtime or driver. Use the exact receipt URL when citing a measured claim.

## Receipts
- RTX 5090 32GB / Llama 3.1 8B: https://bmdpat.com/receipts/rtx-5090/llama-3-1-8b.md (measured_on: 2026-07-09; source: Reports/5090/benchmarks/2026-07-09-5090-sweep-llama8b-gemma26b.csv via /5090-reports/latest.md)
- RTX 5090 32GB / gemma4:26b: https://bmdpat.com/receipts/rtx-5090/gemma4-26b.md (measured_on: 2026-07-09; source: Reports/5090/benchmarks/2026-07-09-5090-sweep-llama8b-gemma26b.csv via /5090-reports/latest.md)
- RTX 3070 8GB / Llama 3.1 8B: https://bmdpat.com/receipts/rtx-3070/llama-3-1-8b.md (measured_on: 2026-07-26; source: Reports/Sizing-Desk/fluarmn-tier-8gb-2026-07-25.jsonl)
- RTX 3070 8GB / Qwen3.5 9B: https://bmdpat.com/receipts/rtx-3070/qwen-3-5-9b.md (measured_on: 2026-07-26; source: Reports/Sizing-Desk/fluarmn-tier-8gb-2026-07-25.jsonl)

## Personalized path
- Choose a GPU and workload: https://bmdpat.com/desk
- Desk method and citation rules: https://bmdpat.com/desk.md