How fast is AI?
Menu

Intelligence & cost · updated Nov 17, 2026

The same intelligence keeps getting cheaper.

Every level of AI capability starts with one lab, one closed model, and a high price. Then the level fills up with competitors and downloadable models until it costs almost nothing. GPT-4-level intelligence is now 520× cheaper than at its debut.

Pick a level of intelligence

Intelligence 125 first arrived in Feb 2024 with Claude 3 Opus (Anthropic), at $30 per million tokens. Today 134 models from 15 labs reach it, 54 of them open-weight. The cheapest, gpt-oss-20b, costs $0.058: 520× cheaper.

First reached

Feb 2024

Claude 3 Opus, Anthropic · launch price $30

Available today

134 models

from 15 labs · 54 open-weight · 173 have ever reached it

Cheapest today

$0.058

gpt-oss-20b · 520× cheaper than at debut

Open-weight lag

4.8 months

Llama 3.1-405B (Meta AI) was the first downloadable model here

Table sorted by standard 3:1 input-to-output token blend.

gpt-oss-20b2025-08OpenAI · USopenECI 135.0$0.058via DeepInfra
gpt-oss-120b2025-08OpenAI · USopenECI 140.2$0.070via DeepInfra
Phi-42024-12Microsoft Research · USopenECI 129.0$0.087via DeepInfra
DeepSeek V4 Flash 07312026-07DeepSeek · ChinaopenECI 154.4$0.090via DeepInfra
Gemma 3 27B2025-03Google DeepMind · USopenECI 127.5$0.10via DeepInfra
Qwen3.5-9B2026-02Alibaba · ChinaopenECI 139.3$0.11via DeepInfra
DeepSeek-V4-Flash2026-04DeepSeek · ChinaopenECI 146.8$0.11via DeepInfra
Qwen3-32B2025-04Alibaba · ChinaopenECI 136.9$0.12via OVHcloud
Qwen3-14B2025-04Alibaba · ChinaopenECI 136.1$0.12via Nebius
Gemini 2.5 Flash-Lite (Jun 2025)2025-06Google DeepMind · USclosedECI 133.1$0.13via Oracle Cloud
GPT-5 nano2025-08OpenAI · USclosedECI 139.1$0.14via Microsoft Azure
Gemma 4 26B A4B2026-04Google DeepMind · USopenECI 141.6$0.14via DeepInfra

Showing the 12 cheapest of 134 available models.

Source: Capability levels use Epoch AI's Capabilities Index (ECI), which stitches many benchmarks into one score anchored at Claude 3.5 Sonnet = 130 and GPT-5 = 150. A model reaches a level when its ECI estimate is at or above it. Prices are providers' list prices per million tokens, blended 3:1 input to output. · Epoch Capabilities Index · LiteLLM price map

Launch price versus today, at every level

Each row is one level of capability. The orange dot is what it cost when the level was new; the blue dot is the cheapest way to buy it today.

$0.010$0.10$1$10$100120GPT-4650×125Claude 3 Opus520×130Claude 3.5 Sonnet100×135o1-mini75×140o1370×145o3190×150GPT-538×155GPT-5.2 Pro290×160GPT-5.4 Pro165Claude Fable 5.12×170Claude Fable 5.21×
Launch price when the level debuted Cheapest price todayUSD per million tokens, blended 3:1 input to output, log scale

The frontier itself stays expensive

What collapses is the price of last year's best. The most capable models today still cost dollars per million tokens:

  • Claude Fable 5.2 · ECI 173.4$16
  • Gemini 4 Carbon · ECI 169.8$4
  • GPT-6 Astra · ECI 169.0$20
  • Claude Opus 5.5 · ECI 168.5$8
  • GPT-6.1 Sol · ECI 168.5$4
  • GPT-6.1 Terra · ECI 167.7$2

How to read this, and its limits

Per token, not per task. Reasoning models spend more tokens per answer, so a per-token price understates their cost per job. Epoch AI's per-question analysis, which accounts for this, still finds the price of a fixed level of performance falling about 13 times a year. The plunging price of thought.

A benchmark score is not a job. Matching a model's capability index does not mean matching it on every task you care about. Treat levels as broad classes, and read the confidence intervals in Epoch's data before splitting hairs.

Launch prices. For models that debuted before this tracker started, launch prices come from archived copies of each lab's own pricing page or launch post. Since then, the sync records each model's first price itself. When a premium tier reached a level first, the debut price is the cheapest option within that level's first month.

Who you buy from matters. The cheapest listing is often an inference host serving an open-weight model, sometimes in a reduced-precision version. List price is not always what large customers pay.