IFX — AI Inference Cost Index

IFX Optimizer · CLI

Find the money you're wasting on AI.

Drop in your codebase. The Optimizer prices every LLM call against the live IFX index and flags the waste — frontier models on trivial tasks, missing caching, silent retries — then proves the savings with your real usage.

We ran it on 22 of our own apps. One was bleeding 32%.

See how it works npx @ifx/optimizer scan .

Free to scan. Source never leaves your machine.

IFX — the benchmark for AI inference costs

IFX Advisor

Find the right AI model.

Live cost data across the leading models, blended into one daily index. Describe your task — the advisor picks the cheapest model that clears the bar.

Get alerted when a cheaper model clears your bar

Cost of Intelligence

Cheapest model meeting each capability bar

Each bar tracks the cheapest blended $/1M-token model whose capability score meets the gate. Base 100 at inception — falling means it is getting cheaper to buy that level of capability.

Scale aa-intelligence-index-v4.1 · base 100 on 2026-06-18

Frontier · ≥50

≈ GPT-5.5 / Opus-class

100.00

0.00%

Champion

Gemini-3.5 Flash

Blended
$3.3750/1M
Capability
50
Qualifiers
5

Capable · ≥40

≈ DeepSeek V4 Pro-class

100.00

0.00%

Champion

GPT-5.4 Mini

Blended
$1.6875/1M
Capability
40
Qualifiers
10

Budget · ≥25

≈ GPT-4-class

100.00

0.00%

Champion

GPT-5.4 Nano

Blended
$0.4625/1M
Capability
38
Qualifiers
20

This Week in Inference Pricing

IFX unchanged at 79.93

The Inference Index closed at 79.93 on August 5, 2026, unchanged from the previous reading. The index basket holds 29 models across leading providers (drawn from a tracked universe of 271), blended at a 3:1 input-to-output token mix.

Per-1M-token prices across the 29 priced basket members today range from $0.06 (lowest blended) to $11.25 (highest blended), with an average blended price of $2.55.

The index is normalized to 100 at inception (2024-04-01), so today's value reads as a percentage of basket cost versus the inception date.

Top movers · 30 day

Biggest price changes across the tracked universe

  • DeepSeek V3.1open-hosted

    DeepSeek

    $0.43/1M-51.43%

  • Llama 4 Scoutopen-hosted

    Meta

    $0.15/1M-46.90%

  • GLM-4.6open-hosted

    Zhipu

    $0.88/1M+18.64%

  • Llama 4 Maverickopen-hosted

    Meta

    $0.35/1M-15.66%