A command-line benchmark for testing AI models with standard prompts across local engines on Apple Silicon Macs. A benchmark is a repeatable test used to measure speed, response time, memory use, temperature, and sometimes power use.

Use it to test coding, tool-use, reasoning, or long text generation; compare two to eight models; measure power; export JSON; view past runs; or create and share a result card.

It removes guesswork when choosing an engine or model for local work. Repeated runs and saved history help you see typical results instead of relying on one test.

bench.ja — Command for coding agents by druide67: A command-line benchmark for testing AI models with standard…

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add commands/druide67/asiai/bench.ja

Clone the repo

git clone --depth 1 https://github.com/druide67/asiai

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/commands/druide67/asiai/bench.ja.svg)](https://agentmods.dev/commands/druide67/asiai/bench.ja)

Your own site

<a href="https://agentmods.dev/commands/druide67/asiai/bench.ja"><img src="https://agentmods.dev/badge/commands/druide67/asiai/bench.ja.svg" alt="Measured on agentmods" height="20"></a>

Per session 43 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 2,144 The whole file, excluding the scripts and references it only reads on demand.

Security scan B 1 finding. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.00043	$0.02144
Opus 5	$0.00022	$0.01072
Sonnet 5	$0.00009	$0.00429
Haiku 4.5	$0.00004	$0.00214

Measured 4d ago against content hash 7910c5ddfe05, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

bench.ja scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Asks for rootmediumPrivilege escalation

A mod that escalates privileges can change anything on the machine, not only the project.

| `--power` | sudo powermetricsで電力をクロスバリデーション（IOReportは常時有効） |

Origin

Copies of this mod

1 near-identical copy found in the catalogue:

bench.ko — 100% identical, 150 lines differ

docs/commands/bench.ja.md · 170 lines

How it starts

The opening of the file, as written. The whole thing — 170 lines — stays where its author put it; the contents beside it link to each section on GitHub.

asiai bench

標準化プロンプトによるクロスエンジンベンチマーク。

使用方法

asiai bench [options]

オプション

オプション	説明
`-m, --model MODEL`	ベンチマーク対象モデル（デフォルト：自動検出）
`-e, --engines LIST`	エンジンフィルター（例：`ollama,lmstudio,mlxlm`）
`-p, --prompts LIST`	プロンプトタイプ：`code`、`tool_call`、`reasoning`、`long_gen`
`-r, --runs N`	プロンプトあたりの実行回数（デフォルト：3、中央値 + 標準偏差用）
`--power`	sudo powermetricsで電力をクロスバリデーション（IOReportは常時有効）
`--context-size SIZE`	コンテキストフィルプロンプト：`4k`、`16k`、`32k`、`64k`
`--export FILE`	結果をJSONファイルにエクスポート
`-H, --history PERIOD`	過去のベンチマークを表示（例：`7d`、`24h`）
`-Q, --quick`	クイックベンチマーク：1プロンプト（code）、1回実行（約15秒）
`--compare MODEL [MODEL...]`	クロスモデル比較（2〜8モデル、`-m`と排他）
`--card`	共有可能なベンチマークカードを生成（ローカルSVG、`--share`でPNG）
`--share`	コミュニティベンチマークデータベースに結果を共有

例

asiai bench -m qwen3.5 --runs 3 --power

  Mac Mini M4 Pro — Apple M4 Pro  RAM: 64.0 GB (42% used)  Pressure: normal

Benchmark: qwen3.5

  Engine       tok/s (±stddev)    Tokens   Duration     TTFT       VRAM    Thermal
  ────────── ───────────────── ───────── ────────── ──────── ────────── ──────────
  lmstudio    72.6 ± 0.0 (stable)   435    6.20s    0.28s        —    nominal
  ollama      30.4 ± 0.1 (stable)   448   15.28s    0.25s   26.0 GB   nominal

  Winner: lmstudio (2.4x faster)
  Power: lmstudio 13.2W (5.52 tok/s/W) — ollama 16.0W (1.89 tok/s/W)

プロンプト

4つの標準化プロンプトが異なる生成パターンをテストします：

名前	トークン	テスト内容
`code`	512	構造化コード生成（PythonでBST）
`tool_call`	256	JSON関数呼び出し / 指示追従
`reasoning`	384	多段階数学問題
`long_gen`	1024	持続スループット（bashスクリプト）

--context-size を使用すると、大規模コンテキストフィルプロンプトでテストできます。

クロスエンジンモデルマッチング

ランナーはエンジン間でモデル名を自動解決します — gemma2:9b（Ollama）と gemma-2-9b（LM Studio）は同じモデルとしてマッチングされます。

JSONエクスポート

結果を共有・分析用にエクスポート：

asiai bench -m qwen3.5 --export bench.json

JSONにはマシンメタデータ、エンジンごとの統計（中央値、CI 95%、P50/P90/P99）、生のランごとのデータ、前方互換性のためのスキーマバージョンが含まれます。

Read the full file on GitHub · 170 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

4d ago First seen · 170 lines · 43 tokens per session scan B 7910c5ddfe05

Subscribe to this mod's changes

bench.ja is a command published in the GitHub repository druide67/asiai (11 stars, last pushed 5d ago), licensed Apache-2.0. It adds 43 tokens to every session and 2,144 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it B with 1 finding (asks for root). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Apple Silicon LLM Writing and docs benchmark cli inference lm-studio

bench.ja

Getting it into your agent

Wrote this? Show the measurements

What it costs to keep this loaded

Grade B, and why

Asks for rootmediumPrivilege escalation

Copies of this mod

How it starts

asiai bench

使用方法

オプション

例

プロンプト

クロスエンジンモデルマッチング

JSONエクスポート

What this file has done since we first saw it

Other commands, from other repositories

perfup

integrate-pipeline

netllm-connect

netllm-setup

netllm-swarm

/opsx-explore