Data as of 2 Nov 2026 · last updated · by BenchLeader
The median ranked model on BenchLeader costs $0.50 per 1M input tokens and $2.00 per 1M output tokens (239 ranked models with a list price). Blended at three input tokens to one output token, the median is $0.85 per 1M tokens, and output costs a median 4× input. The cheapest model scoring 65+ on the BenchLeader Index is GPT-6.1 Terra at $1.00/$5.00.
Want your own number? The LLM cost calculator prices a month of your usage on every ranked model at these list prices.
The median ranked model on BenchLeader costs $0.50 per 1M input tokens and $2.00 per 1M output tokens (239 ranked models with a list price). Each dot is a ranked model; named dots are the cheapest in each Index tier. Chart image (SVG)
Key statistics
Each statistic has its own link (#) to cite it directly.
The median ranked model costs $0.50 per 1M input tokens and $2.00 per 1M output tokens, across 239 ranked models with a list price (BenchLeader, 2 Nov 2026). #
The median blended price (three input tokens to one output token) is $0.85 per 1M tokens (BenchLeader, 2 Nov 2026). #
Output tokens cost a median 4× as much as input tokens (BenchLeader, 2 Nov 2026). #
The cheapest ranked model scoring 65+ on the BenchLeader Index is GPT-6.1 Terra by OpenAI, at $1.00 input and $5.00 output per 1M tokens; the tier's median is $8.00 blended (BenchLeader, 2 Nov 2026). #
The cheapest ranked model scoring 60–65 on the BenchLeader Index is Gemini 3.7 Flash by Google, at $0.75 input and $3.75 output per 1M tokens; the tier's median is $4.00 blended (BenchLeader, 2 Nov 2026). #
The cheapest ranked model scoring 55–60 on the BenchLeader Index is GPT-5 nano by OpenAI, at $0.05 input and $0.40 output per 1M tokens; the tier's median is $1.35 blended (BenchLeader, 2 Nov 2026). #
Among ranked models scoring 60–65, blended prices run from $1.50 to $67.50 per 1M tokens, a 45× spread for similar Index scores (BenchLeader, 2 Nov 2026). #
3 ranked models are on an announced launch price: Gemini 4 Argon until 30 Nov 2026, Gemini 3.6 Flash until 31 Dec 2026 and Gemini 3.8 Flash until 31 Dec 2026 (BenchLeader, 2 Nov 2026). #
Cheapest model at each quality tier
Ranked models grouped by BenchLeader Index. Blended price assumes three input tokens for every output token.
Current list prices of the ranked models released each month. This is a snapshot of today's prices grouped by release date, not a price history: models launched in different months differ in size and purpose.
Released
Ranked models with a price
Median input per 1M
Median output per 1M
January 2026 · small sample
3
$0.15
$0.60
February 2026
17
$0.31
$2.40
March 2026
10
$0.40
$2.00
April 2026
16
$0.975
$3.30
May 2026 · small sample
7
$1.50
$7.50
June 2026 · small sample
7
$0.95
$4.00
July 2026
13
$1.00
$4.05
August 2026
16
$0.555
$2.24
September 2026
15
$2.00
$6.00
October 2026
15
$0.57
$1.44
Launch prices with an end date
Ranked models whose provider has announced the current price ends. The figures above use today's price.
239 ranked models (those with a BenchLeader Index) have a list price; prices are per million tokens, at each model's best-scoring configuration. The blended price weights input three to one, the mix of a typical chat or agent turn. Cached input, batch discounts and coding plans are not counted. Full sourcing is in the methodology.
This page is a snapshot of today's list prices; how the price of each quality level has fallen since mid-2024 is on LLM price trends. For the long run, Epoch AI estimates the cost of reaching a given level of performance has fallen about 47% per quarter since 2023 (The plunging price of thought).
Cite these statistics
List prices are each provider's published rates, read via Artificial Analysis, LiteLLM price table, OpenRouter, models.dev. They are shown here with attribution and are not offered for reuse; BenchLeader's own figures on this page are the Index scores and the medians. Cite the page below.
Citation
BenchLeader, “LLM API price statistics”, data as of 2 Nov 2026, https://www.benchleader.com/stats/llm-prices.
Questions people ask
How much does an LLM API cost per million tokens?
The median ranked model on BenchLeader costs $0.50 per 1M input tokens and $2.00 per 1M output tokens (239 ranked models with a list price). Blended at three input tokens to one output token, the median is $0.85.
What is the cheapest high-quality LLM?
GPT-6.1 Terra by OpenAI is the cheapest ranked model scoring 65+ on the BenchLeader Index, at $1.00 input and $5.00 output per 1M tokens.
Why do output tokens cost more than input tokens?
Generating a token takes a full pass of the model for each token, while input is processed in parallel. Across ranked models, output costs a median 4× input.
Are LLM prices falling?
This page is today's list prices. For how the price of a given quality level has fallen since mid-2024, see LLM price trends (/stats/llm-price-trends). Over the longer run, Epoch AI estimates that the cost of reaching a given level of performance has fallen about 47% per quarter since 2023 (“The plunging price of thought”).
