GPT-4o-mini Cost Optimization: 4 Tested Ways to Save 50%

1. GPT-4o-mini Current Market Price (2026-06-19)

Pricing on OpenAI Official: $0.15/M input, $0.60/M output. But relays like Aihubmix and API2D can reach $0.105/M input — a 30% gap. See the full cross-provider comparison on the pricing page.

2. 4 Tested Ways to Save 50%

1. Method 1: Pick a high-coverage relay — the lowest price among verification ≥60% on the pricing page = $0.105/M, saving 30% vs official
2. Method 2: Use prompt cache — GPT-4o-mini supports OpenAI caching discounts; long prompts (≥1k tokens) can save 50%
3. Method 3: Prefer short prompts — input price is 1/4 of output price, so shorter prompts are more cost-effective (aggregation/classification tasks)
4. Method 4: Streaming output — streaming responses = client can interrupt = save wasted output (20-40% on long answers)

3. How to Use Cache Tiers (See L3 Details)

The GPT-4o-mini prompt cache discount is OpenAI's official pricing: $0.075/M cache read. This means the second call with the same prompt gets the input price halved. See tier details in the GPT-4o-mini Official L3 Details (including cache read 5m / 1h tiers).

4. Decision Checklist

1. Check the current public price on OpenAI Official = cost floor

2. Check Aihubmix GPT-4o-mini Details = lowest relay price

3. Run a test with OpenAI Detection = protocol passthrough verification