docs: remove PR benchmark page

[ofir-frd]

No description provided.

[qodo-free-for-open-source-projects]

Review Summary by Qodo

Remove PR benchmark documentation and navigation

📝 Documentation

Walkthroughs

Description

• Remove PR benchmark documentation page entirely
• Delete benchmark results table and model analysis
• Remove benchmark navigation from docs sidebar
• Clean up changelog reference to benchmark

Diagram

flowchart LR
  A["PR Benchmark Docs"] -->|delete| B["Removed from docs/"]
  C["Navigation Entry"] -->|remove| D["Removed from mkdocs.yml"]
  E["Changelog Reference"] -->|delete| F["Removed from README.md"]
  G["Summary Page Link"] -->|remove| H["Removed from summary.md"]

Loading

File Changes

1. docs/docs/pr_benchmark/index.md 📝 Documentation +0/-505

Delete entire PR benchmark documentation file

• Entire 505-line benchmark documentation file deleted
• Removed methodology section explaining benchmark approach
• Deleted comprehensive results table with 24 LLM models
• Removed detailed analysis sections for each model variant
• Deleted appendix with example results and model list

docs/docs/pr_benchmark/index.md

---

2. docs/mkdocs.yml 📝 Documentation +0/-3

Remove PR benchmark from navigation sidebar

• Removed PR Benchmark navigation section from sidebar
• Deleted reference to pr_benchmark/index.md in nav structure
• Cleaned up navigation hierarchy

docs/mkdocs.yml

---

3. docs/docs/summary.md 📝 Documentation +0/-4

Remove PR benchmark link from summary

• Removed PR Benchmark section from summary page
• Deleted link to pr_benchmark/index.md
• Cleaned up trailing whitespace

docs/docs/summary.md

---

View more (1)

4. README.md 📝 Documentation +0/-2

Remove benchmark reference from changelog

• Removed changelog entry referencing PR benchmark results
• Deleted link to benchmark page from GPT-5 announcement
• Cleaned up commented changelog section

README.md

---

[qodo-free-for-open-source-projects]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (1) 📎 Requirement gaps (0)

1. Missing newline in summary.md 📘 Rule violation ✓ Correctness

Description

docs/docs/summary.md now has no trailing newline at EOF, which is a formatting disruption that can
cause noisy diffs and tooling inconsistencies. Add a final newline to preserve existing file
formatting conventions.

Code

docs/docs/summary.md[45]

* [Self-reflection](core-abilities/self_reflection.md)
-
-## PR Benchmark
-
-* [PR Benchmark](pr_benchmark/index.md)

Evidence

PR Compliance ID 9 requires preserving existing file formatting conventions and avoiding unnecessary
formatting churn; the diff explicitly indicates the file now lacks a newline at end of file.

AGENTS.md
docs/docs/summary.md[43-45]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`docs/docs/summary.md` is missing a trailing newline at EOF, which disrupts existing formatting conventions.

## Issue Context
The PR diff shows `\ No newline at end of file` after removing the PR Benchmark section.

## Fix Focus Areas
- docs/docs/summary.md[43-45]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

ⓘ The new review experience is currently in Beta. Learn more

[PeterDaveHello]

Hi @ofir-frd @naorpeled,
The benchmark was really useful for comparing pr-agent across different models. Since it's been removed, is there still any evals data or model comparison info available?
Happy to contribute resources if it helps.

[ofir-frd]

While a complete revamp of Qodo docs is underway and is planned to include a benchmark section, the following blogs provide this data:

https://www.qodo.ai/blog/qodo-outperforms-claude-in-code-review-benchmark/

https://www.qodo.ai/blog/nvidia-nemotron-3-super-is-closing-the-gap-for-open-source-models/

https://www.qodo.ai/blog/how-we-built-a-real-world-benchmark-for-ai-code-review/

Please note that the benchmark here was for Qodo-Merge and not for PR-Agent, and the blogs refer to the new Qodo 2.x app.

We are in the process of clarifying the separation between Qodo's community-powered project support in PR-Agent and the company's products.

[PeterDaveHello]

Hi @ofir-frd,

Thanks for the quick reply, the blog links, and for clarifying that those benchmarks are for Qodo-Merge and the new 2.x app rather than PR-Agent.

I've been following the latest updates and recently became a subscriber. It would still be very helpful to have some evaluation or model comparison data specifically for PR-Agent.

Happy to help test or contribute if you plan to add it in the future.