# openrouter-benchmarks

This skill queries OpenRouter's Benchmarks API and pulls model rankings from Artificial Analysis and Design Arena. Handy when you need to pick a model for a task with evidence behind the choice.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, OpenCode · 325 installs · ★ 306
- Author: OpenRouterTeam
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/OpenRouterTeam/skills/tree/HEAD/skills/openrouter-benchmarks/SKILL.md
- Updated: 2026-10-24
- HTML: https://affpapa.org/en/nevibecode/skills/openrouter-benchmarks

## What it does

The skill calls `GET /api/v1/benchmarks` and returns rows of model scores. Artificial Analysis supplies `intelligence_index`, `coding_index` and `agentic_index`, where higher is better. Design Arena gives `elo`, `win_rate`, generation time, arena and category. Filtering stays simple: `source`, `task_type` with `coding`, `intelligence` or `agentic`, plus `arena` and `category` for Design Arena. There is also `max_results`.

Set `OPENROUTER_API_KEY` first — any valid OpenRouter key works, and benchmarks do not need a management key. The skill does not hand you a winner right away. It checks availability through `GET /api/v1/models` and provider endpoints, because a benchmark `model_permaslug` is not always the routable model ID. When no exact `id` match exists but a model carries a matching `canonical_slug`, the row counts as evidence for that family rather than a ready recommendation. Candidates with degraded endpoints or zero uptime drop out of the primary list.

The two sources use different scales, so the skill keeps them apart instead of building one absolute leaderboard — compare within a source. For creative writing and editing there is no direct prose-quality benchmark here; `intelligence_index` is only a weak general-capability signal. Citation metadata `meta.citation`, `meta.source_url` and `meta.as_of` is preserved when results are republished.

## Install

### Claude Code (native platform)

1. **Skills CLI (Vercel)**
```bash
npx skills add OpenRouterTeam/skills --skill openrouter-benchmarks -a claude-code
```

2. **Manually: put the files in place**
- File: `~/.claude/skills/openrouter-benchmarks/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/openrouter-benchmarks/

### Codex

1. **Skills CLI (Vercel)**
```bash
npx skills add OpenRouterTeam/skills --skill openrouter-benchmarks -a codex
```

2. **Manually: put the files in place**
- File: `.agents/skills/openrouter-benchmarks/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

1. **Skills CLI (Vercel)**
```bash
npx skills add OpenRouterTeam/skills --skill openrouter-benchmarks -a cursor
```

2. **Manually: put the files in place**
- File: `.cursor/skills/openrouter-benchmarks/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install OpenRouterTeam/skills openrouter-benchmarks
```

2. **Manually: put the files in place**
- File: `.github/skills/openrouter-benchmarks/SKILL.md`

### Gemini CLI

1. **Skills CLI (Vercel)**
```bash
npx skills add OpenRouterTeam/skills --skill openrouter-benchmarks -a gemini-cli
```

2. **Platform CLI**
```bash
gemini skills install https://github.com/OpenRouterTeam/skills --path skills/openrouter-benchmarks
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/openrouter-benchmarks/SKILL.md` — The legacy .windsurf/skills/ path still works

### OpenCode

**Skills CLI (Vercel)**
```bash
npx skills add OpenRouterTeam/skills --skill openrouter-benchmarks -a opencode
```

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/OpenRouterTeam/skills/tree/HEAD/skills/openrouter-benchmarks/SKILL.md

## When to use it
- pick a model for a coding task from scores instead of chat hearsay
- compare models within Artificial Analysis or Design Arena
- verify a model is actually available on OpenRouter before recommending it
- collect citations and the as-of date for a benchmark report
- rule out a ranking leader whose endpoints are unusable

## Example prompts
- show me the top 10 coding models from Artificial Analysis
- which models lead Design Arena in the uicomponent category
- find a model for an agentic workflow and check it is available on OpenRouter

## Similar
- [pp-openrouter](https://affpapa.org/en/nevibecode/skills/pp-openrouter.md) — skill · 322 installs
- [nano-banana-pro-openrouter](https://affpapa.org/en/nevibecode/skills/nano-banana-pro-openrouter.md) — skill · 9910 installs
- [omni-models](https://affpapa.org/en/nevibecode/skills/omni-models.md) — skill · 1000 installs
- [openrouter](https://affpapa.org/en/nevibecode/skills/openrouter.md) — skill · 403 installs
- [qwencloud-model-selector](https://affpapa.org/en/nevibecode/skills/qwencloud-model-selector.md) — skill · 1600 installs
- [modelchk](https://affpapa.org/en/nevibecode/skills/modelchk.md) — skill · 1200 installs

From the same set: [openrouterteam/skills](https://affpapa.org/en/nevibecode/by/openrouterteam/skills.md) — 4 cards

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md