Which LLM is the cheapest for our workload?

The best-for pages rank the live catalogue per task by the index that measures it, and the router picks the cheapest model that meets your quality rule for each request.

How much can we save on our LLM bill?

On the public benchmark the router saved 89% against sending everything to the reference model. Import last month's provider bill on the bill page and the estimate is computed on your own traffic before you change a line of code.

How often do model prices change?

The catalogue is refreshed from the vendors' price lists; each model page shows the date of its data.

Automatic model selection

Automatic model selection that reads the prompt, not a config file.

Task, complexity, domain, language and sensitivity are read from each prompt; every model you allow is scored on quality for that task, cost for that length and latency; your rules remove the ones that may not answer. The best value is chosen per request and explained - no model names in your code.

Get a free API key Route this request in the playground

Free plan, no card. Open source under Apache-2.0; self-hosting is free forever.

From the benchmark

9.53 vs 9.15

judged quality, selected vs always-frontier

On 150 real prompts two blind judges scored the automatically selected answers higher than the frontier model's, while the selection spent 89% less. The categories where it lost are listed on the benchmark page.

The whole benchmark, with the cases the router lost

The problem

A model chosen once is the wrong model for most requests.

The requests an application receives are not alike: a one-line rewrite, a forty-page analysis, a code fix, a question with a customer's account number in it. One model for all of them is either too expensive for most or too weak for some - and a list of regexes in the application is not a selection strategy.

Simple requests pay frontier prices
Hard requests fail quietly on the cheap model
Sensitive requests go wherever the default points

How the router answers it

Three things it does for every request.

Signals from the prompt
Task type, complexity, domain, language, whether it carries personal data or code - extracted on the way in, in microseconds, before any model is called.
A score per candidate
Published benchmarks, declared capabilities, a small routing model trained on outcomes and your feedback give each candidate a quality estimate for this task; cost and latency complete the score.
Your weights, your floor
Balanced, Cheapest, Best or Fastest - or your own weights - and a quality floor below which a candidate is not considered. A low-confidence decision can fall back to the model you name.

Watch one decision

Illustrative: the lowest cost that still clears the quality the task needs. The live quote below is the real one.

A real request

Our churn rose from 3.1% to 5.4% while NPS stayed flat. Reason step by step about the causes.

The decision names the target, its confidence and the ranked alternatives with their prices. Replace the prompt with a one-line rewrite and the same call picks a small, fast model.

Route it in the playground and read the trace

The callbash

curl -s https://opensmartroute.ai/api/v1/route \
  -H "Authorization: Bearer $OSR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text": "<prompt>"}'

Where this stands next to the alternatives, and how to go deeper.

Install

npm install @opensmartroute/sdk

Typed SDKs for Python and TypeScript

Or no install at all: the hosted API answers a plain curl with the decision, and the browser extension shows the price and the router's pick under the composer of the AI chat sites you already use.

For the business

What it means for which models you pay for.

Catalogue of every vendor
The right model, not the famous one
Price, context, abilities and benchmark indexes side by side, with the model that fits each task named.
The rankings
89% less
Lower cost per request
On the 150-prompt benchmark the router spent 89% less than sending every prompt to the reference model, and the judged quality went up, not down.
Read the benchmark
Per million tokens, every model
Prices you can compare
The catalogue prints input and output prices and what three typical monthly workloads cost on each.
The model catalogue

Which LLM is the cheapest for our workload?: The best-for pages rank the live catalogue per task by the index that measures it, and the router picks the cheapest model that meets your quality rule for each request.
How much can we save on our LLM bill?: On the public benchmark the router saved 89% against sending everything to the reference model. Import last month's provider bill on the bill page and the estimate is computed on your own traffic before you change a line of code.
How often do model prices change?: The catalogue is refreshed from the vendors' price lists; each model page shows the date of its data.

Route your first request in under a minute.

OpenSmartRoute is open source and the free plan keeps the full trace of every decision. Create a workspace, mint a key and send the request above.

Create a free account Read the benchmark

Free plan, no card. Fifteen thousand decisions a month with the full trace.

A model chosen once is the wrong model for most requests.

Automatic model selection that reads the prompt, not a config file.

A model chosen once is the wrong model for most requests.

Three things it does for every request.

Signals from the prompt

A score per candidate

Your weights, your floor

Our churn rose from 3.1% to 5.4% while NPS stayed flat. Reason step by step about the causes.

Where this stands next to the alternatives, and how to go deeper.

What it means for which models you pay for.

The right model, not the famous one

Lower cost per request

Prices you can compare

Route your first request in under a minute.

Automatic model selection that reads the prompt, not a config file.

A model chosen once is the wrong model for most requests.

Three things it does for every request.

Signals from the prompt

A score per candidate

Your weights, your floor

Our churn rose from 3.1% to 5.4% while NPS stayed flat. Reason step by step about the causes.

Where this stands next to the alternatives, and how to go deeper.

What it means for which models you pay for.

The right model, not the famous one

Lower cost per request

Prices you can compare

Route your first request in under a minute.