Status: [LIVE]

Endpoints found:
https://developers.openai.com/api/docs/pricing

Source: https://github.com/BUTTERGANG/ai-pricing-research

# AI Pricing Research

Historical and current AI model pricing data — LLM API costs, market landscape, free models, and provider comparisons. Built as a local knowledge base for trend analysis and sentiment correlation.

## Structure

- `kb/price_points_*.jsonl` — canonical structured datasets (one JSONL file per provider):
  - `price_points_openai.jsonl` (37 entries, 2020-2026)
  - `price_points_anthropic.jsonl` (27 entries, 2023-2026)
  - `price_points_google.jsonl` (31 entries, 2023-2026)
  - `price_points_deepseek.jsonl` (9 entries, 2023-2026)
  - `price_points_xai.jsonl` (5 entries, 2025-2026)
- `kb/market_data.md` — market-level context (sizing, spend, infrastructure, adoption)
- `kb/PRICE_POINTS_SPEC.md` — data contract and field definitions
- `kb/INDEX.md` — source registry and retrieval metadata
- `kb/CONVENTIONS.md` — project-level accuracy instructions
- `reports/` — formatted markdown reports derived from the dataset
- `notes/` — source digests and extraction notes
- `scripts/` — fetchers, parsers, loaders

## Current totals

- **216 price points** across 11 providers (OpenAI 37, Anthropic 27, Google 31, DeepSeek 11, xAI 5, Mistral 26, Cohere 13, Meta/Llama hosted 11, Alibaba/Qwen 25, Moonshot/Kimi 17, OpenRouter free 13)
- **17 sources** catalogued (provider pricing pages, aggregators, market analysts)
- Time span: **2020 through projected 2027** (Gemini 3.8 Flash Jan 2027 rate)

## Quick start

```bash
cd ~/code/BUTTERGANG/ai-pricing-research

# Inspect the dataset
wc -l kb/price_points_*.jsonl
head -5 kb/price_points_openai.jsonl

# Convert to CSV for analysis
python3 -c "
import json, csv, glob
with open('kb/all_price_points.csv', 'w', newline='') as out:
    w = csv.writer(out)
    w.writerow(['id','provider','model_name','effective_date','input_price','output_price','price_type','context_window','source_url'])
    for f in sorted(glob.glob('kb/price_points_*.jsonl')):
        for line in open(f):
            d = json.loads(line)