Canonical: https://botbento.com/usecases/review-new-inference-benchmark-runs-against-a/
Format: Markdown representation of the public HTML page.

# Review new inference benchmark runs against a stored baseline

[Operations](https://botbento.com/usecases/ops/) · Intermediate

Spot throughput and latency changes before they reach the team

A recurring review bot that checks newly added benchmark outputs against a saved baseline, highlights notable throughput or latency changes, and prepares a concise regression note for a maintainer to review before anything is shared.

## Quick answers

**What does it do?** Spot throughput and latency changes before they reach the team. Result: Draft regression summary for approval.

**What does it need?** Tools: GitHub, Docs, Notes. Intermediate: some setup.

**How do I run it?** On a schedule: Weekly benchmark review. Copy the prompt below into a bot in BotBento.

## How it works

1. On a schedule: Weekly benchmark review
2. Then: Compare latest benchmark run to baseline
3. Result: Draft regression summary for approval

## Prompt to start with

```text
Each week, read the newest benchmark result files in the repository and compare key throughput and latency numbers with the stored baseline. Flag any metric that moved in an unexpected direction. Draft a short summary listing the changed metrics and the likely cause. Do not post, send, or open issues; present the draft for approval first.
```

## Tools it uses

GitHub, Docs, Notes

## Works with

BotBento

## Tags

benchmarks, performance, review, inference

## Source

[Show HN: TrnRT – An open-source LLM inference engine for AWS Trainium](https://news.ycombinator.com/item?id=50533270) — Hacker News · checked 2026-11-17

Summarized in our own words. Any results described there are the authors’ own, not measured by BotBento.

Verified 2026-11-17: passed every automated gate and a second-model review (kimi-code/k3.1-pro). Automated, not human review. [How verification works](https://botbento.com/usecases/verification/)

By BotBento Editorial · Added 2026-11-17 · Updated 2026-11-17 · [Suggest a correction](https://botbento.com/contact/)

## Related use cases

- [Fraud rule drift review and variant selection](https://botbento.com/usecases/fraud-rule-drift-review-and-variant-selection/) — Check rule performance, spot drift and propose the best variant
- [Contract review bot that flags terms before anything is signed](https://botbento.com/usecases/contract-review-bot-that-flags-terms-before/) — Reads incoming contract PDFs and drafts a plain-language brief for review
- [Vet third-party AI tool packages before installation](https://botbento.com/usecases/vet-third-party-ai-tool-packages-before/) — Compare declared capabilities with actual files and probe risky servers.
