modelparams

Typed model parameters for Python, generated from the open modelparams.dev catalog.

pip install modelparams

Typed request parameters

Generated TypedDict definitions provide autocomplete and static errors for unsupported keys, incorrect value types, and invalid enum values:

from modelparams import validate_params
from modelparams.types.openai import Gpt_4_1Params

params: Gpt_4_1Params = {
    "max_tokens": 1024,
    "temperature": 0.7,
}

validated = validate_params("openai/gpt-4.1", params)

validate_params uses a cached strict Pydantic adapter. It returns a provider-keyed dictionary or raises pydantic.ValidationError for unknown parameters, coercions, invalid enum values, or values outside the catalog range.

from openai import OpenAI

OpenAI().chat.completions.create(model="gpt-4.1", messages=messages, **validated)

Provider-specific dot paths remain literal dictionary keys:

from modelparams.types.anthropic import Claude_Haiku_4_5_20251001Params

params: Claude_Haiku_4_5_20251001Params = {
    "thinking.type": "enabled",
    "thinking.budget_tokens": 4096,
}

Catalog helpers

from modelparams import get_defaults, get_model, get_param, list_models

model = get_model("anthropic/claude-haiku-4-5-20251001")
print(model.auth_type, model.params)

defaults = get_defaults("anthropic/claude-haiku-4-5-20251001")
thinking = get_param("anthropic/claude-haiku-4-5-20251001", "thinking.type")
anthropic_models = list_models("anthropic")

The catalog is bundled with the package. No network request is made at runtime.

Development

From the repository root:

npm run codegen:python
uv sync --project packages/modelparams-python --extra dev
uv run --project packages/modelparams-python pytest packages/modelparams-python/tests

Generated catalog and type files are committed and verified in CI. Python releases use independent modelparams-py@x.y.z tags and publish to PyPI through the repository's trusted-publisher workflow.

Catalog entries expose api_surface as a typed value, so callers can distinguish request families such as openai-chat-completions and openai-responses before using a parameter set.

Metadata

Release files for modelparams 0.0.48

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for modelparams 0.0.48
File	Size	Uploaded
modelparams-0.0.48.tar.gz	51.0 kB	Sep 30, 2026	Details

Built distribution (wheel)

Table of built distributions (wheels) for modelparams 0.0.48
File	Interpreter	ABI	Platform	Reset
modelparams-0.0.48-py3-none-any.whl 68.4 kB Sep 30, 2026	Python 3	none	any	Details

Total release size: 119.4 kB

Release files / modelparams-0.0.48.tar.gz

Download URL	modelparams-0.0.48.tar.gz
Size	51.0 kB
Tags	Source
SHA-256 checksum How to use checksums	`7874684e315da8806a5802198f8ef2be5dbc94879a4fed0938cdaa9d26bb7387`
BLAKE2b-256 checksum How to use checksums	`74ed241335b01839d6cd1a6bf8a2ee94bc86d650581a571f4e3eae16c5897089`
Upload date	Sep 30, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/mnfst/modelparams.dev
Publishing commit: github.com/mnfst/modelparams.dev/tree/2df6840d67ef98693bf75f25d972e73efa060a29

Workflow

Publishing configuration: github.com/mnfst/modelparams.dev/blob/2df6840d67ef98693bf75f25d972e73efa060a29/.github/workflows/release-modelparams-python.yml
Publishing logs: github.com/mnfst/modelparams.dev/actions/runs/36674715110/attempts/1

Artifact

Subject: modelparams-0.0.48.tar.gz
SHA-256 checksum: 7874684e315da8806a5802198f8ef2be5dbc94879a4fed0938cdaa9d26bb7387 Verifying attestations

Release files / modelparams-0.0.48-py3-none-any.whl

Download URL	modelparams-0.0.48-py3-none-any.whl
Size	68.4 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`0c44f4b0649d9806aaa9821fa38d19b4b08b64361123831a5648c0956428d62f`
BLAKE2b-256 checksum How to use checksums	`6da905a06d44c8a6873ba26ea120b3ba75f1510b838b88f4a5469df928c8243a`
Upload date	Sep 30, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Provenance

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/mnfst/modelparams.dev
Publishing commit: github.com/mnfst/modelparams.dev/tree/2df6840d67ef98693bf75f25d972e73efa060a29

Workflow

Publishing configuration: github.com/mnfst/modelparams.dev/blob/2df6840d67ef98693bf75f25d972e73efa060a29/.github/workflows/release-modelparams-python.yml
Publishing logs: github.com/mnfst/modelparams.dev/actions/runs/36674715110/attempts/1

Artifact

Subject: modelparams-0.0.48-py3-none-any.whl
SHA-256 checksum: 0c44f4b0649d9806aaa9821fa38d19b4b08b64361123831a5648c0956428d62f Verifying attestations