modelparams.dev
Sampling number 111 models

Frequency penalty frequency_penalty

Penalizes tokens in proportion to how often they have already appeared.

Type
number
Default
0
Range
-2 – 2

frequency_penalty by model

The exact type, default, range, and conditions for frequency_penalty on each model that documents it.

Model Provider Type Default Range / values Condition
Zai GLM-4.7 Cerebras number 0 -2…2 step 0.1 —
Command A 03 2025 Cohere number 0 0…1 step 0.1 —
Command A Plus 05 2026 Cohere number 0 0…1 step 0.1 —
Command A Reasoning 08 2025 Cohere number 0 0…1 step 0.1 —
Command A Translate 08 2025 Cohere number 0 0…1 step 0.1 —
Command A Vision 07 2025 Cohere number 0 0…1 step 0.1 —
Command R 08 2024 Cohere number 0 0…1 step 0.1 —
Command R Plus 08 2024 Cohere number 0 0…1 step 0.1 —
Command R7B 12 2024 Cohere number 0 0…1 step 0.1 —
DeepSeek v4 Flash 0731 Fireworks number — -2…2 step 0.1 —
DeepSeek v4 Flash Vision Exp Fireworks number — -2…2 step 0.1 —
DeepSeek v4 Pro Fireworks number — -2…2 step 0.1 —
DeepSeek v4 Pro 0813 Fireworks number — -2…2 step 0.1 —
DeepSeek V4p1 Flash Fireworks number — -2…2 step 0.1 —
Ember 1 Fireworks number — -2…2 step 0.1 —
GLM-5p2 Fireworks number — -2…2 step 0.1 —
GLM-5p3 Fireworks number — -2…2 step 0.1 —
GLM-5p3 Flash Fireworks number — -2…2 step 0.1 —
GPT-OSS 120B Fireworks number — -2…2 step 0.1 —
GPT-OSS 20B Fireworks number — -2…2 step 0.1 —
Inkling Fireworks number — -2…2 step 0.1 —
Kimi K2p6 Fireworks number — -2…2 step 0.1 —
Kimi K2p7 Code Fireworks number — -2…2 step 0.1 —
Kimi K3 Fireworks number — -2…2 step 0.1 —
MiniMax M2p7 Fireworks number — -2…2 step 0.1 —
MiniMax M3 Fireworks number — -2…2 step 0.1 —
Muse Glimmer 30B Fireworks number — -2…2 step 0.1 —
Nemotron 3 Ultra Nvfp4 Fireworks number — -2…2 step 0.1 —
Nemotron Lightning 3p5 30B A3B Fireworks number — -2…2 step 0.1 —
Qwen3p7 Plus Fireworks number — -2…2 step 0.1 —
Qwen3p8 2p4t A95B Fireworks number — -2…2 step 0.1 —
Qwen3p8 Max Fireworks number — -2…2 step 0.1 —
Allam 2 7B Groq number 0 -2…2 step 0.1 —
GPT-OSS 120B Groq number 0 -2…2 step 0.1 —
GPT-OSS 20B Groq number 0 -2…2 step 0.1 —
GPT-OSS Safeguard 20B Groq number 0 -2…2 step 0.1 —
Qwen3 32B Groq number 0 -2…2 step 0.1 —
Qwen3.6 27B Groq number 0 -2…2 step 0.1 —
Qwen3.8 27B Groq number 0 -2…2 step 0.1 —
Muse Spark 1.1 Meta number 0 -2…2 step 0.1 —
Muse Spark 1.2 Contributor Meta number 0 -2…2 step 0.1 —
Codestral 2508 Mistral number 0 -2…2 step 0.1 —
Codestral Latest Mistral number 0 -2…2 step 0.1 —
Devstral 2512 Mistral number 0 -2…2 step 0.1 —
Devstral Latest Mistral number 0 -2…2 step 0.1 —
GLM-5.2 Mistral number 0 -2…2 step 0.1 —
Magistral Medium Latest Mistral number 0 -2…2 step 0.1 —
Magistral Small Latest Mistral number 0 -2…2 step 0.1 —
Ministral 14B 2512 Mistral number 0 -2…2 step 0.1 —
Ministral 14B Latest Mistral number 0 -2…2 step 0.1 —
Ministral 3B 2512 Mistral number 0 -2…2 step 0.1 —
Ministral 3B Latest Mistral number 0 -2…2 step 0.1 —
Ministral 8B 2512 Mistral number 0 -2…2 step 0.1 —
Ministral 8B Latest Mistral number 0 -2…2 step 0.1 —
Mistral Code Latest Mistral number 0 -2…2 step 0.1 —
Mistral Large 2512 Mistral number 0 -2…2 step 0.1 —
Mistral Large Latest Mistral number 0 -2…2 step 0.1 —
Mistral Medium Mistral number 0 -2…2 step 0.1 —
Mistral Medium 2604 Mistral number 0 -2…2 step 0.1 —
Mistral Medium 3 Mistral number 0 -2…2 step 0.1 —
Mistral Medium 3.5 Mistral number 0 -2…2 step 0.1 —
Mistral Medium Latest Mistral number 0 -2…2 step 0.1 —
Mistral NeMo Mistral number 0 -2…2 step 0.1 —
Mistral Small 2603 Mistral number 0 -2…2 step 0.1 —
Mistral Small Latest Mistral number 0 -2…2 step 0.1 —
Voxtral Small 2507 Mistral number 0 -2…2 step 0.1 —
Voxtral Small Latest Mistral number 0 -2…2 step 0.1 —
Zai GLM-5 Mistral number 0 -2…2 step 0.1 —
Zai GLM-5.2 Mistral number 0 -2…2 step 0.1 —
Zai GLM-5.3 Mistral number 0 -2…2 step 0.1 —
Zai GLM-Latest Mistral number 0 -2…2 step 0.1 —
Moonshot v1 128K Moonshot AI number 0 -2…2 step 0.1 —
Moonshot v1 32K Moonshot AI number 0 -2…2 step 0.1 —
Moonshot v1 8K Moonshot AI number 0 -2…2 step 0.1 —
DeepSeek v4 Flash 0731 NVIDIA number 0 -2…2 —
DeepSeek v4 Pro 0813 NVIDIA number 0 -2…2 —
Diffusiongemma 26B A4B IT NVIDIA number 0 -2…2 —
Gemma 4 31B IT NVIDIA number 0 -2…2 —
GLM-5.3 Flash NVIDIA number 0 -2…2 —
GPT-OSS 120B NVIDIA number 0 -2…2 —
GPT-OSS 20B NVIDIA number 0 -2…2 —
Kimi K3 NVIDIA number 0 -2…2 —
Laguna Xs 2.1 NVIDIA number 0 -2…2 —
Llama 3.1 NemoGuard 8B Topic Control NVIDIA number 0 -2…2 —
Llama 3.1 Nemotron Nano 8B v1 NVIDIA number 0 -2…2 —
Llama 3.1 Nemotron Ultra 253B v1 NVIDIA number 0 -2…2 —
Llama 3.2 11B Vision Instruct NVIDIA number 0 -2…2 —
Llama 3.2 90B Vision Instruct NVIDIA number 0 -2…2 —
Llama 3.3 Nemotron Super 49B v1 NVIDIA number 0 -2…2 —
Llama 3.3 Nemotron Super 49B v1.5 NVIDIA number 0 -2…2 —
MiniMax M3 NVIDIA number 0 -2…2 —
Mistral Nemotron NVIDIA number 0 -2…2 —
Muse Glimmer 30B NVIDIA number 0 -2…2 —
Nemotron Mini 4B Instruct NVIDIA number 0 -2…2 —
Riva Translate 4B Instruct v1.1 NVIDIA number 0 -2…2 —
DeepSeek Flash Subscription Opencode Go number — -2…2 step 0.1 —
DeepSeek v4 Flash Vision Exp Subscription Opencode Go number — -2…2 step 0.1 —
GLM-5.3 Subscription Opencode Go number — -2…2 step 0.1 —
GLM-5.3 Flash Subscription Opencode Go number — -2…2 step 0.1 —
Hy3 Subscription Opencode Go number — -2…2 step 0.1 —
Hy4 Preview Subscription Opencode Go number — -2…2 step 0.1 —
Longcat 2.0 Subscription Opencode Go number — -2…2 step 0.1 —
Longcat 2.5 Preview Free Subscription Opencode Go number — -2…2 step 0.1 —
MiMo v2.6 Flash Subscription Opencode Go number — -2…2 step 0.1 —
MiMo v2.6 Pro Subscription Opencode Go number — -2…2 step 0.1 —
MiniMax M2.7 Subscription Opencode Go number — -2…2 step 0.1 —
Omen Alpha Subscription Opencode Go number — -2…2 step 0.1 —
Qwen3.8 Flash Subscription Opencode Go number — -2…2 step 0.1 —
Qwen3.8 Max Subscription Opencode Go number — -2…2 step 0.1 —
Space Bunny Free Subscription Opencode Go number — -2…2 step 0.1 —
MiMo v2.5 Pro Xiaomi number 0 -2…2 step 0.1 —

Models without frequency_penalty documented yet

These models don't list frequency_penalty in the catalog. That may mean the provider doesn't accept it, or it simply hasn't been added yet — the data is community-maintained.

Opencode Go

Thinking Machines

Related sampling parameters

Resources

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/models/openai/gpt-5.5.json
curl https://modelparams.dev/api/v1/models/openai/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