> What AI capability costs per million tokens, which model is cheapest at each level of capability, and how fast the price of a given capability has fallen.

147 models with a public price and a score

# Cost

What capability costs today, what a task really costs once thinking is counted, and how fast each level gets cheaper.

## Price against capability · list prices · log scale

Current versions

All

_Chart: Index score against blended price per million tokens_

Price is blended 3:1 input to output tokens. The line traces the best score you can buy at each price: anything under it is beaten by something cheaper. Dashed rings are estimated scores.

## What a task really costs · per Artificial Analysis test task · current models

Cheapest first

Strongest first

DeepSeek-V4.1-Flash lists 5.8× cheaper per token than Inkling-Small, yet costs 3.0× more per task on Artificial Analysis's test set. The bill depends on how many tokens a model uses, not just its rate.

Model

Score

Cost per task

Token price

LongCat 2.0

~140

$0.060

$0.52/M

MiMo-V2.6-Flash

~167

$0.060

$0.18/M

GPT-6 Luna

166

$0.070

$0.20/M

TM

Inkling-Small

154

$0.090

$0.64/M

gpt-oss-120b

131

$0.11

$0.070/M

Gemini 3.5 Flash-Lite

142

$0.12

$0.85/M

MiMo-V2.6-Pro

~178

$0.13

$0.54/M

Qwen3.5-9B

130

$0.21

$0.11/M

Qwen3.7-Plus

147

$0.22

$0.56/M

GLM-5.3-Flash

158

$0.25

$0.24/M

DeepSeek-V4.1-Flash

165

$0.27

$0.11/M

Qwen3.8-Flash-Next

~169

$0.37

–

Qwen3.5

146

$0.47

$1.29/M

Qwen3.6-35B-A3B

139

$0.48

$0.36/M

Mistral Medium 3.5

134

$0.50

$3.0/M

MiniMax-M3

145

$0.51

$0.52/M

Kimi K2.7 Code

153

$0.54

$1.34/M

Nemotron 3 Ultra

145

$0.60

$1.05/M

Gemini 3.1 Pro

166

$0.67

$4.5/M

DeepSeek-V4-Pro

167

$0.67

$1.98/M

GPT-5.5 Instant

136

$0.69

–

Step 5 Preview

~175

$0.72

–

GPT-6.1 Sol

184

$0.72

$4.0/M

Qwen3.8-27B

152

$1.01

$1.06/M

Gemini 3.8 Flash

171

$1.24

$1.5/M

GPT-5.6 Terra

178

$1.4

$4.5/M

Muse Spark 1.3

170

$1.6

$2.0/M

Nemotron 3 Super

~132

$1.64

$0.17/M

Gemini 4 Argon

~188

$1.99

–

Kimi K3

172

$2.0

$3.01/M

GLM-5.3

167

$2.01

$1.33/M

Qwen3.8-2.4T-A95B

~169

$2.16

$3.0/M

GPT-6 Astra

197

$3.26

$20/M

Grok 4.7

~178

$3.74

$3.0/M

Qwen3.8-Max

170

$5.41

–

Claude Opus 5.5

200

$5.98

$8.0/M

Claude Sonnet 5.5

193

$7.62

$4.0/M

Claude Fable 5.1

192

$7.63

$20/M

Average cost of one task in Artificial Analysis's evaluation suite, at the setting shown on hover, counting every token the model uses (input, reasoning and answer). Token price is OpenRouter's list price, blended 3:1, for reference.

## The cheapest model at each level, by release date · today's list prices · log scale

_Chart: Cheapest model at each capability level over time_

Each line steps down when a cheaper model reaching that score is released, priced at what it costs today. It shows how quickly cheaper models arrive, not what anyone paid at the time (GPT-4 itself listed at $37.50 per million tokens in March 2023).

---
Source: https://themodelindex.org/cost/ · The Model Index · data as of 2026-10-01