Artificial Analysis APIartificialanalysis.ai ↗
Access AI model benchmarks, provider leaderboards, GPU hardware stats, and pricing data from Artificial Analysis via 7 structured endpoints.
What is the Artificial Analysis API?
This API surfaces 7 endpoints covering AI model benchmarks, provider leaderboards, and hardware performance data from Artificial Analysis. The get_llm_providers_leaderboard endpoint returns the full dataset of provider-model combinations with pricing, latency, throughput, and benchmark scores across every tracked API host. Additional endpoints cover model details by slug, GPU chip specifications, and speech, image, and video model families.
No input parameters required.
curl -X GET 'https://api.parse.bot/scraper/ae371dc0-d495-4644-ae2f-97fbf3214b85/get_llm_providers_leaderboard' \ -H 'X-API-Key: $PARSE_API_KEY'
Typed, relational, agent-ready
A generated client with real types, enums, and the links between objects — the structure a flat JSON response can't carry. Autocompletes in your editor and reads cleanly to coding agents.
- Fully typed · autocompletes
- Objects link to objects
- Typed errors & pagination
Typed Python client. Set up the SDK in your uv project, then pull this API’s typed client:
uv add parse-sdk uv run parse init uv run parse add --marketplace artificialanalysis-ai-api
uv run parse add --marketplace pulls a pinned snapshot of this canonical API — it won’t change underneath you. To customize it, subscribe and swap to your own copy.
"""Walkthrough: Artificial Analysis SDK — compare AI models, providers, and hardware."""
from parse_apis.Artificial_Analysis_API import (
ArtificialAnalysis, Query, ModelNotFound
)
client = ArtificialAnalysis()
# Search for GPT models and print their summaries.
for model in client.models.search(query=Query.GPT, limit=5):
print(model.name, model.slug, model.is_reasoning)
# Drill into one model via constructible shortcut, then list its providers.
detail = client.model("o3")
for provider in detail.providers.list(limit=3):
print(provider.slug, provider.price_1m_input_tokens, provider.price_1m_output_tokens)
# Browse the LLM providers leaderboard.
for endpoint in client.provider_endpoints.list(limit=3):
print(endpoint.name, endpoint.price_1m_input_tokens, endpoint.context_window_tokens)
# Get hardware benchmark data.
for bench in client.system_benchmarks.list(limit=3):
print(bench.label, bench.precision, bench.inference_framework_name)
# Typed error handling: attempt to fetch a model that does not exist.
try:
client.models.get(slug="nonexistent-model-xyz")
except ModelNotFound as exc:
print(f"Model not found: {exc.slug}")
print("exercised: models.search / model() constructible / providers.list / provider_endpoints.list / system_benchmarks.list")
Get the main LLM API Providers Leaderboard data, including all model endpoints and their performance metrics (speed, latency, price, etc.). Returns a large dataset of provider-model combinations with pricing, timescale performance data, and benchmark scores.
No input parameters required.
{
"type": "object",
"fields": {
"endpoints": "array of provider-model endpoint objects with pricing, performance metrics, host info, model benchmarks, and timescale data"
}
}About the Artificial Analysis API
LLM Provider Leaderboard and Model Data
get_llm_providers_leaderboard returns the complete provider-model matrix tracked by Artificial Analysis. Each entry includes pricing (input/output token costs), latency and throughput metrics across timescales, and benchmark scores. get_models_list returns a flat array of all tracked LLM models with fields like slug, name, shortName, isReasoning, deprecated, and nested creator info. To look up a specific model, pass its slug to get_model_detail — for example gpt-4o, o3, or gpt-oss-20b — and get back the full model object. get_model_providers uses the same slug input and extends the response with a providers array that includes per-provider pricing, host_api_id, cache pricing, and feature support flags.
Hardware Benchmarks
get_hardware_benchmarks returns two arrays: chips (unique GPU chip objects with memory bandwidth, capacity, power draw, and creator info) and system_benchmarks (per-system measurements including throughput, output speed, and latency alongside the inference framework used). This lets you compare raw hardware performance across GPU configurations independently of any specific model provider.
Search and Multimodal Models
search_models accepts a query string and performs case-insensitive substring matching against name, slug, and shortName fields across the full model list — useful for quickly narrowing down candidates like claude, llama, or gpt. get_speech_image_video_models returns four separate arrays for image generation, video generation, speech-to-text, and text-to-speech model families, each with name, slug, url, and creator metadata.
The Artificial Analysis API is a managed, monitored endpoint for artificialanalysis.ai — not a raw scraper you maintain. Every endpoint is automatically health-checked on a schedule, and when artificialanalysis.ai changes and a check fails, the API is automatically queued for repair and re-verified. It is built to keep working as the site underneath it changes.
This isn't an official artificialanalysis.ai API — it's an independent, maintained REST wrapper over public data. Where the source has no official API (or only a limited one), Parse gives you a stable contract over a source that never promised one, and keeps it current. Need a new endpoint or field? You can revise it yourself in plain English and the agent rebuilds it against the live site in minutes — contributing the change back to the shared API is free.
Will this API break when the source site changes?+
Is this an official API from the source site?+
Can I fix or extend this API myself if I need a new endpoint or field?+
What happens if I call an endpoint that has an issue?+
- Build a model selection tool that ranks LLM providers by price-per-token and latency using leaderboard endpoint data
- Track which models are flagged as reasoning-capable via the
isReasoningfield returned byget_models_list - Compare GPU hardware throughput and memory bandwidth across chips using
get_hardware_benchmarksfor infrastructure planning - Look up all API providers offering a specific model slug and their associated token pricing via
get_model_providers - Aggregate benchmark scores across model families to produce custom LLM comparison dashboards
- Discover available speech-to-text and text-to-speech model families for voice application planning via
get_speech_image_video_models - Search and filter the tracked model catalog by name substring to automate model discovery pipelines
| Tier | Price | Credits/month | Rate limit |
|---|---|---|---|
| Free | $0/mo | 100 | 5 req/min |
| Hobby | $30/mo | 1,000 | 20 req/min |
| Developer | $100/mo | 5,000 | 100 req/min |
One credit = one API call regardless of which marketplace API you call. Exceeding the rate limit returns a 429 response. Authenticate with the X-API-Key header.
Does Artificial Analysis have an official developer API?+
What does `get_model_providers` return beyond basic model info?+
get_model_providers returns a model object with full benchmark scores, evaluation scores, host models, and creator info, plus a providers array where each entry includes input/output token pricing, cache pricing, host_api_id, and feature support flags for that specific provider endpoint. This makes it the right endpoint when you need to compare all providers hosting a single model.Are there any known limitations with model slug lookups?+
get_model_detail and get_model_providers can return upstream errors for certain slugs if the corresponding model page on Artificial Analysis has incomplete data. It's worth validating slugs first via get_models_list before passing them to detail endpoints.