菜单
Query Artificial Analysis API to compare LLM benchmarks, pricing, and speed. Use when exploring which AI model to use for a task, comparing model performance, or checking pricing/speed tradeoffs.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when working with NobodyWho local LLM inference docs, APIs, model selection, GGUF model loading, chat, streaming, tool calling, structured output, embeddings/RAG, or NobodyWho integrations for Python, Kotlin, Swift, React Native, Flutter, or Godot.
Primary W&B skill for broad or mixed Weights & Biases work: project overviews, W&B runs and artifacts, Weave traces and evaluations, Reports, Signal Builder, and Launch workflows. Use when the task spans multiple W&B surfaces or the user asks generally what is happening in a W&B project.
Use when the user mentions Verda Cloud, GPU/CPU VMs, cloud instances, deploying servers, ML training infrastructure, cloud costs/billing, SSH into remote machines, object storage / S3 buckets, uploading or downloading files, or verda CLI commands.
Send a macOS notification when the user asks to be notified, alerted, pinged, or reminded that something is done.
Interact with local Chrome browser session (only on explicit user approval after being asked to inspect, debug, or interact with a page open in Chrome)
Use this skill to search flights using google flights
| name | artificial-analysis |
| description | Query Artificial Analysis API to compare LLM benchmarks, pricing, and speed. Use when exploring which AI model to use for a task, comparing model performance, or checking pricing/speed tradeoffs. |
Query the Artificial Analysis free API to compare AI models on benchmarks, pricing, and speed. Useful when deciding which LLM (or image/video/speech model) to use for a specific task.
Requires an API key set as ARTIFICIAL_ANALYSIS_API_KEY environment variable.
set -Ux ARTIFICIAL_ANALYSIS_API_KEY "your-key")Full docs (endpoints, fields, response formats): https://artificialanalysis.ai/api-reference#free-api
Rate limit: 1,000 requests/day. The rate limit is strict and resets slowly — avoid burst calls. Always cache responses to a temp file (e.g., /tmp/aa_models.json) and reuse them instead of making repeated API calls. Only fetch fresh data if no cached file exists or the user explicitly asks to refresh.
curl -s "https://artificialanalysis.ai/api/v2/data/llms/models" \
-H "x-api-key: $ARTIFICIAL_ANALYSIS_API_KEY" | python3 -m json.tool
Key response fields: evaluations (benchmark scores), pricing (per 1M tokens), median_output_tokens_per_second, median_time_to_first_token_seconds.
curl -s "https://artificialanalysis.ai/api/v2/data/media/text-to-image" \
-H "x-api-key: $ARTIFICIAL_ANALYSIS_API_KEY"
Add ?include_categories=true for per-category ELO breakdown.
curl -s "https://artificialanalysis.ai/api/v2/data/media/image-editing" \
-H "x-api-key: $ARTIFICIAL_ANALYSIS_API_KEY"
curl -s "https://artificialanalysis.ai/api/v2/data/media/text-to-speech" \
-H "x-api-key: $ARTIFICIAL_ANALYSIS_API_KEY"
curl -s "https://artificialanalysis.ai/api/v2/data/media/text-to-video" \
-H "x-api-key: $ARTIFICIAL_ANALYSIS_API_KEY"
Add ?include_categories=true for per-category breakdown.
curl -s "https://artificialanalysis.ai/api/v2/data/media/image-to-video" \
-H "x-api-key: $ARTIFICIAL_ANALYSIS_API_KEY"
Add ?include_categories=true for per-category breakdown.
The free API does not include openness/licensing data. When filtering for open-weight models, use web searches to verify which models are actually open before presenting results.
Use jq to filter/sort results (e.g., | jq '.data | sort_by(.pricing.price_1m_blended_3_to_1)')
For LLMs, compare evaluations.artificial_analysis_intelligence_index against pricing.price_1m_blended_3_to_1 for value analysis
Use model id fields as stable identifiers — name and slug may change
Attribution required: link to https://artificialanalysis.ai/ when sharing data