Is the API free to use?

Yes, the API is free for non-commercial use with attribution. Commercial use requires written permission from DemandSphere.

How often is the data updated?

The data is updated when new frontier models launch or existing models receive significant updates to pricing, benchmarks, or capabilities. The updated field in the response shows the last build date.

What license applies to this data?

CC BY-NC 4.0 (Creative Commons Attribution-NonCommercial 4.0 International). You must credit DemandSphere and link back to the tracker. No commercial use without permission.

What data is included?

40 frontier AI models across 14 providers with benchmarks (MMLU-Pro, GPQA Diamond, SWE-bench, HumanEval, LiveCodeBench, MATH, AIME, HLE), API pricing, context windows, parameter counts, capabilities, and provider metadata.

Tracker API

Free JSON endpoint for frontier AI model data. Benchmarks, pricing, and capabilities across 42 models and 14 providers.

Endpoint

        GET https://www.demandsphere.com/research/ai-frontier-model-tracker/api.json
        
      

Returns the full dataset as a single JSON object. No authentication required. No rate limiting. Response is static and cached by Cloudflare.

License

CC BY-NC 4.0 View license

Attribution required - credit DemandSphere and link to the tracker
Non-commercial use only
Commercial use requires written permission
Contact: demandsphere.com/contact

Response Format

Top-level fields

          license: "CC BY-NC 4.0"

          license_url: string

          attribution: "DemandSphere AI Frontier Model Tracker"

          attribution_url: string

          terms: string

          updated: "YYYY-MM-DD"

          model_count: number

          provider_count: number

          providers: { ... }

          models: [ ... ]

Model object fields

          id: string

          name: string

          prov: string (provider key)

          type: "General" | "Reasoning"

          acc: "Open" | "Closed"

          rel: "YYYY-MM" (release date)

          par: number | null (total params B)

          act: number | null (active params B)

          ctx: number (context window in K)

          mm: boolean (multimodal)

          r: boolean (reasoning/CoT)

          inp: number | null ($/M input tokens)

          out: number | null ($/M output tokens)

          cite: string (source URL)

          desc: string (description)

Benchmark Fields

All benchmark values are percentages (0-100) or null if not published. Null means the provider did not report this benchmark, not that the model scored zero.

mmlu

MMLU-Pro - graduate-level knowledge

gpqa

GPQA Diamond - PhD-level science

swe

SWE-bench Verified - GitHub issue resolution

he

HumanEval pass@1 - code generation

lcb

LiveCodeBench - interactive coding

math

MATH Level-5 / MATH-500

aime

AIME 2025 - competition math

hle

Humanity's Last Exam - ceiling test

Example Usage

curl

curl https://www.demandsphere.com/research/ai-frontier-model-tracker/api.json

curl + jq

curl -s https://www.demandsphere.com/research/ai-frontier-model-tracker/api.json \
  | jq '.models | sort_by(-.gpqa) | .[0:5] | .[] | {name, gpqa, prov}'

JavaScript

fetch('https://www.demandsphere.com/research/ai-frontier-model-tracker/api.json')
  .then(r => r.json())
  .then(data => console.log(data.models.length + ' models'));

Python

import requests
data = requests.get('https://www.demandsphere.com/research/ai-frontier-model-tracker/api.json').json()
for m in sorted(data['models'], key=lambda x: x.get('gpqa') or 0, reverse=True)[:5]:
    print(f"{m['name']}: {m.get('gpqa', 'N/A')}% GPQA")

Required Attribution

When using this data, include one of the following:

      <!-- HTML -->

      Data from <a href="https://www.demandsphere.com/research/ai-frontier-model-tracker/">DemandSphere AI Frontier Model Tracker</a>

      # Markdown / plain text

      Data: DemandSphere AI Frontier Model Tracker (demandsphere.com/research/ai-frontier-model-tracker)

Track how these models see your brand

Monitor your brand visibility across ChatGPT, Gemini, AI Overviews, AI Mode, Perplexity, Copilot, and more.

Learn more Get a demo