modelparams.dev

Model parameters vs. API parameters

Two unrelated things go by the name "parameters", and people looking for one routinely land on the other. The short version: a parameter count tells you how big a model is, and this site doesn't track it. API parameters are what you put in the request body, and that's the entire catalog.

Parameter count: the weights

A model's parameter count is how many trained weights it has. GPT-3 shipped with 175 billion of them. Llama 3.1 405B puts the number in its name. It's a rough proxy for how expensive a model is to run, and not much else — a good 70B model from this year will beat a 175B model from 2020 on almost any task you care about.

For most of the models people ask about, the number isn't public. OpenAI has never published a count for GPT-3.5 or anything since. Neither has Anthropic for Claude, or Google for Gemini. Figures that circulate for those models are estimates and leaks. Open-weight models are the exception: if you can download the weights you can count them, which is why Llama, Qwen, DeepSeek and Mistral all state exact numbers in their model cards.

API parameters: the request settings

These are the fields you send when you call a model: temperature, top_p, max_tokens, stop, reasoning effort, and whatever else a given endpoint accepts. They change what comes back for a given prompt. They don't change the model.

Every provider names and gates them differently. OpenAI's temperature range isn't Google's. Some parameters only apply once another one is set, and some quietly stop working on reasoning models. That's the mess this catalog documents: 242 models under one schema, with the type, default, valid range, and gating conditions for each field.

Which one are you after?

The request settings

Defaults, ranges and gating conditions for every tracked model.

Browse the catalog → Read the parameter glossary →

The weight count

We don't have it. Check the model card on Hugging Face for open-weight models, or the provider's own announcement post. For closed models, expect to find nothing official.

Frequently asked questions

Does modelparams.dev list how many parameters a model has?
No. 242 models are in the catalog and none of them carry a weight count. We track API parameters, meaning the settings you send in a request. For an open-weight model the count is usually in its Hugging Face model card. For a closed model there is often no published figure at all.
How many parameters does GPT-3.5 have?
OpenAI has never published it. The 175 billion figure people repeat belongs to GPT-3, which is a different model, and every number quoted for GPT-3.5 and later is an estimate rather than a disclosure.
Are model parameters and hyperparameters the same thing?
No, and that is a third meaning of the word. Hyperparameters are the settings used to train a model, like learning rate and batch size. Parameters are the weights that training produces. API parameters are what you send at inference time. Same word, three unrelated things.
Is temperature a model parameter?
Temperature is an API parameter. It changes how the output is sampled and is not stored in the model. Two requests to the same model at different temperatures run against identical weights.

How to use

Building with an AI agent? Hit Copy to grab this whole guide as Markdown and paste it in — or point your agent straight at /llms.txt.

modelparams.dev is an open, community-maintained catalog of model parameters. Each entry shows the knobs you can turn — type, default, range, and the conditions that gate it.

The same model accessed via an API key and via a subscription usually exposes a different set of parameters. We list both as separate entries so the data stays honest.

Catalog API

The full catalog is static JSON, CORS-enabled, served from the edge.

curl https://modelparams.dev/api/v1/models.json

Each entry is keyed by provider/model for API-key variants; subscription variants append -subscription.

If you only need the params for one model contract, use the providerless endpoint. Subscription contracts are model slugs with -subscription.

curl https://modelparams.dev/api/v1/params/gpt-5.5.json
curl https://modelparams.dev/api/v1/params/gpt-5.5-subscription.json

Single model

curl https://modelparams.dev/api/v1/models/anthropic/claude-opus-4-7.json
curl https://modelparams.dev/api/v1/models/anthropic/claude-opus-4-7-subscription.json

JSON Schema

Every entry validates against a JSON Schema you can use in your editor or pipeline.

curl https://modelparams.dev/api/v1/schema.json

Add this header to any YAML you author for autocomplete in VS Code:

# yaml-language-server: $schema=https://modelparams.dev/api/v1/schema.json

Logos

Provider logos are available at /assets/logos/{provider}.svg where {provider} is the provider slug. They use currentColor so they inherit your text color.

curl https://modelparams.dev/assets/logos/anthropic.svg

Logos are sourced from the models.dev repo (MIT) and used under nominative fair use.

Contribute

The data lives in YAML under models/{provider}/{model}-{auth}.yaml in the GitHub repo. Open a PR; CI validates against the schema and rebuilds.

Edit on GitHub MIT licensed