Compare AI models,
task by task.
Tick models to see them side by side: price, context size, features and benchmark scores (up to 4 at a time). Free models are highlighted. Every figure has a source and a date.
- Models in the catalog
- 338
- Free models
- 13
List entries: 345. A free route of a model is its own entry, next to the paid model.
Showing 121–130 of 130
How to read this list
Tick models to see them side by side: price, context size, features and benchmark scores (up to 4 at a time). Free models are highlighted. Every figure has a source and a date.
List entries: 345. A free route of a model is its own entry, next to the paid model.
Tips for paying less for AI All free models
Mixed price per 1M tokens: 3 input tokens to 1 output token.
Free: no charge for tokens on that route. Limits and data terms are set by the provider.
The large price is the mixed price. Input and output prices are shown under it.
A label next to a task (Top 10% or Top 25%) means a third party measured the model and it ranks in that share of the scored models. The scores themselves are on the model page and on the task page.
The share is rounded up to whole models, so a rank slightly past the exact share still counts. Models with equal scores share the better rank.
"On OpenRouter since" is the date the model appeared in the OpenRouter list. The vendor's own release date can differ.
Models that think before they answer can use many more output tokens, so your real cost can be higher than the figure shown.
Websites and apps · About this task
Models
-
gpt-oss-120b
Context 131k · Accepts text · Tools · Reasoning · Weights on Hugging Face
$0.0703 mixed / 1M tokens in $0.037 · out $0.17 via DeepInfra -
Qwen3 30B A3B
Context 41k · Accepts text · Tools · Reasoning · Weights on Hugging Face
$0.215 mixed / 1M tokens in $0.12 · out $0.50 via DeepInfra -
Mistral Small 3.2 24B
Context 128k · Accepts text and images · Tools · Weights on Hugging Face
$0.1063 mixed / 1M tokens in $0.075 · out $0.20 via DeepInfra -
Llama 4 Maverick
Context 524k · Accepts text and images · Tools · Weights on Hugging Face
$0.5125 mixed / 1M tokens in $0.35 · out $1.00 via Parasail -
gpt-oss-20b
Context 131k · Accepts text · Tools · Reasoning · Weights on Hugging Face
$0.034 mixed / 1M tokens in $0.017 · out $0.085 via Darkbloom -
Nova Premier 1.0
Context 1M · Accepts text and images · Tools
No data -
GPT-4o
Context 128k · Accepts text, images, and files · Tools
$4.38 mixed / 1M tokens in $2.50 · out $10.00 -
Nova Pro 1.0
Context 300k · Accepts text and images · Tools
$1.40 mixed / 1M tokens in $0.80 · out $3.20 -
Llama 4 Scout
Context 328k · Accepts text and images · Tools · Weights on Hugging Face
$0.15 mixed / 1M tokens in $0.10 · out $0.30 via DeepInfra -
GPT-5.2-Codex
Context 400k · Accepts text and images · Tools · Reasoning
$4.81 mixed / 1M tokens in $1.75 · out $14.00 via Azure
No model matches these filters.
Try fewer filters or another word.
Method and sources
- Where the numbers come from
- Models, features and prices: OpenRouter. Second opinion on vendor prices: LiteLLM. Exchange rates: NBP. Benchmark scores: Artificial Analysis and Design Arena, relayed by OpenRouter.
- How tasks are assigned
- A task chip is shown only when a third party has measured the model and it ranks in the top 10% or the top 25% of the scored models. Tasks matched by a declared feature (context size, input and output types) are listed on the task page and the model page, without a chip. We do not score models ourselves.
- Cheaper alternatives
- On a model page we list models with an Intelligence Index close to or above the model's, a clearly lower price and the same input types. It is a rule applied to published numbers, not a recommendation.
- Free models
- A free model is a route that charges nothing per token. A free route often has different limits from the paid version of the same model, so it is a separate listing that reuses the benchmark scores of the paid version.
- Artificial Analysis / Design Arena
- Intelligence, Coding and Agentic indices: Artificial Analysis (artificialanalysis.ai). Design categories: Design Arena (designarena.ai). Both are relayed by the OpenRouter public model list and were read on November 21, 2026. We compute ranks, tiers and costs from these numbers; the sources have not reviewed them. Scores belong to the model, not to the route it is served on. Ranks are among all models we list, including older ones. The sources do not publish how many comparisons or runs stand behind each number; new models can still move.
Artificial Analysis · Design Arena · Relayed by OpenRouter. opens in a new tab
- Fixed numbers we chose
- Top 10% and Top 25% bands, at least 30 scored models for a band label, long documents from 400k tokens of context, budget up to $0.30 mixed price per 1M tokens, "ending soon" within 30 days.
- Exchange rates
- Exchange rate, NBP: November 20, 2026