Githubusercontent APIraw.githubusercontent.com ↗
Query WebLLM's prebuilt model list at any git ref. Returns model IDs, WASM URLs, VRAM requirements, and Hugging Face links filtered by substring.
What is the Githubusercontent API?
The get_model_configs endpoint reads the WebLLM (mlc-ai/web-llm) prebuilt model list from any git ref and returns up to 7 structured fields per matching model entry, including resolved WASM URLs and VRAM estimates. Supply one or more comma-separated substrings via model_id_contains to filter models by name, and set ref to pin a specific release tag or branch such as v0.2.80 or main.
curl -X GET 'https://api.parse.bot/scraper/3569befc-5b12-416f-abdb-41d0cc500a83/get_model_configs?ref=main&model_id_contains=SmolLM2-360M-Instruct-q4f32_1%2CSmolLM2-1.7B-Instruct-q4f32_1' \ -H 'X-API-Key: $PARSE_API_KEY'
Typed, relational, agent-ready
A generated client with real types, enums, and the links between objects — the structure a flat JSON response can't carry. Autocompletes in your editor and reads cleanly to coding agents.
- Fully typed · autocompletes
- Objects link to objects
- Typed errors & pagination
Typed Python client. Set up the SDK in your uv project, then pull this API’s typed client:
uv add parse-sdk uv run parse init uv run parse add --marketplace raw-githubusercontent-com-api
uv run parse add --marketplace pulls a pinned snapshot of this canonical API — it won’t change underneath you. To customize it, subscribe and swap to your own copy.
"""Walkthrough: WebLLM prebuilt model config lookup — bounded, re-runnable."""
from parse_apis.raw_githubusercontent_com_api import WebLlm, InputNotFound
client = WebLlm()
# Search for specific model configs by substring match against model_id.
for cfg in client.model_configs.search(
model_id_contains="SmolLM2-360M-Instruct-q4f32_1,SmolLM2-1.7B-Instruct-q4f32_1",
limit=5,
):
print(cfg.model_id, cfg.vram_required_mb, "MB VRAM")
print(" HF:", cfg.model)
print(" context window:", cfg.overrides.context_window_size)
print(" low resource:", cfg.low_resource_required)
# Drill into a single result with .first().
hit = client.model_configs.search(
model_id_contains="SmolLM2-360M-Instruct-q4f32_1",
limit=1,
).first()
if hit is not None:
print("WASM lib:", hit.model_lib)
print("required features:", hit.required_features)
print("matched on:", hit.matched_substrings)
# A non-existent git ref raises InputNotFound.
try:
client.model_configs.search(
model_id_contains="SmolLM2-360M-Instruct-q4f32_1",
ref="v0.2.80",
limit=1,
).first()
except InputNotFound as e:
print("expected error for bad ref:", e.message)
print("exercised: model_configs.search")
Fetches src/config.ts from the WebLLM repository at the given git ref, parses the prebuiltAppConfig.model_list, and returns one record per entry whose model_id contains at least one of the requested substrings. Each record carries the Hugging Face model URL, model_id, the fully resolved model_lib WASM URL (prefix and version constants from the same file are substituted), vram_required_MB, low_resource_required, required_features, overrides, and which requested substrings matched. The identifiers array confirms, per requested substring, whether any model_id contained it. A single round trip; total_models_in_list is the size of the full list at that ref. A substring that matches nothing yields present=false and no record (valid, not an error). A ref with no src/config.ts (for example tag v0.2.80, which does not exist in the repository) returns a 422 input_not_found error.
| Param | Type | Description |
|---|---|---|
| ref | string | Git branch, tag, or commit of mlc-ai/web-llm to read, e.g. main or a release tag such as v0.2.81. |
| model_id_containsrequired | string | Comma-separated list of substrings to match against model_id (case-sensitive), e.g. SmolLM2-360M-Instruct-q4f32_1,SmolLM2-1.7B-Instruct-q4f32_1. The scraper splits on commas; do NOT pass an array. At most 50 substrings are honored. |
{
"type": "object",
"fields": {
"ref": "git ref that was read",
"model_list": "array of matching entries: model (HF URL), model_id, model_lib (resolved WASM URL), vram_required_MB (number), low_resource_required (boolean|null), required_features (array of strings), overrides (object), matched_substrings (array)",
"identifiers": "array of {substring, present(boolean)} confirming each requested substring",
"model_version": "modelVersion constant from config.ts",
"model_lib_url_prefix": "modelLibURLPrefix constant from config.ts",
"total_models_in_list": "integer count of all entries in model_list at this ref"
},
"sample": {
"data": {
"ref": "main",
"model_list": [
{
"model": "https://huggingface.co/mlc-ai/SmolLM2-1.7B-Instruct-q4f32_1-MLC",
"model_id": "SmolLM2-1.7B-Instruct-q4f32_1-MLC",
"model_lib": "https://raw.githubusercontent.com/mlc-ai/binary-mlc-llm-libs/main/web-llm-models/v0_2_86/base/SmolLM2-1.7B-Instruct-q4f32_1_cs1k-webgpu.wasm",
"overrides": {
"context_window_size": 4096
},
"vram_required_MB": 2692.38,
"required_features": [],
"matched_substrings": [
"SmolLM2-1.7B-Instruct-q4f32_1"
],
"low_resource_required": true
},
{
"model": "https://huggingface.co/mlc-ai/SmolLM2-360M-Instruct-q4f32_1-MLC",
"model_id": "SmolLM2-360M-Instruct-q4f32_1-MLC",
"model_lib": "https://raw.githubusercontent.com/mlc-ai/binary-mlc-llm-libs/main/web-llm-models/v0_2_86/base/SmolLM2-360M-Instruct-q4f32_1_cs1k-webgpu.wasm",
"overrides": {
"context_window_size": 4096
},
"vram_required_MB": 579.61,
"required_features": [],
"matched_substrings": [
"SmolLM2-360M-Instruct-q4f32_1"
],
"low_resource_required": true
}
],
"identifiers": [
{
"present": true,
"substring": "SmolLM2-360M-Instruct-q4f32_1"
},
{
"present": true,
"substring": "SmolLM2-1.7B-Instruct-q4f32_1"
}
],
"model_version": "v0_2_86/base",
"model_lib_url_prefix": "https://raw.githubusercontent.com/mlc-ai/binary-mlc-llm-libs/main/web-llm-models/",
"total_models_in_list": 167
},
"status": "success"
}
}About the Githubusercontent API
What the API Returns
The single get_model_configs endpoint parses the prebuiltAppConfig.model_list from WebLLM's src/config.ts at a caller-specified git ref and returns one record per entry whose model_id contains at least one of the requested substrings. Each record in model_list includes the Hugging Face model URL (model), the model_id string, a fully resolved WASM library URL (model_lib), vram_required_MB as a number, and low_resource_required as a boolean. The response also carries top-level constants model_version and model_lib_url_prefix extracted from the same file.
Filtering and Ref Selection
The model_id_contains parameter accepts a comma-separated list of substrings and is case-sensitive — SmolLM2 and smollm2 are treated differently. The response includes an identifiers array that echoes each requested substring alongside a present boolean, so callers can confirm which substrings matched anything at all. The optional ref parameter defaults to a known stable tag but accepts any valid git branch, tag, or full commit SHA from the mlc-ai/web-llm repository. total_models_in_list reports the count of all entries in the list at that ref, regardless of how many matched the filter.
Practical Uses
This API is useful for tooling that needs to know whether a specific quantized model variant (e.g., a q4f16 or q0f32 build) is available in a given WebLLM release, or to retrieve the correct WASM URL for a model before bundling it into a web application. Because model_lib is fully resolved — not a relative path — the returned URL can be used directly in WebLLM's runtime configuration without further string manipulation.
The Githubusercontent API is a managed, monitored endpoint for raw.githubusercontent.com — not a raw scraper you maintain. Every endpoint is automatically health-checked on a schedule, and when raw.githubusercontent.com changes and a check fails, the API is automatically queued for repair and re-verified. It is built to keep working as the site underneath it changes.
This isn't an official raw.githubusercontent.com API — it's an independent, maintained REST wrapper over public data. Where the source has no official API (or only a limited one), Parse gives you a stable contract over a source that never promised one, and keeps it current. Need a new endpoint or field? You can revise it yourself in plain English and the agent rebuilds it against the live site in minutes — contributing the change back to the shared API is free.
Will this API break when the source site changes?+
Is this an official API from the source site?+
Can I fix or extend this API myself if I need a new endpoint or field?+
What happens if I call an endpoint that has an issue?+
- Verify whether a quantized model variant exists in a specific WebLLM release before shipping a web app
- Retrieve the resolved WASM library URL for a model to embed directly in a WebLLM runtime config
- Check VRAM requirements across model variants to gate model selection based on device capability
- Audit which models were added or removed between two git refs by comparing total_models_in_list
- Build a model picker UI that shows Hugging Face links and VRAM estimates for browser-side LLM deployment
- Confirm low_resource_required flags to filter models appropriate for mobile or constrained devices
| Tier | Price | Credits/month | Rate limit |
|---|---|---|---|
| Free | $0/mo | 200 | 5 req/min |
| Hobby | $30/mo | 1,000 | 20 req/min |
| Developer | $100/mo | 5,000 | 100 req/min |
| Team | $300/mo | 20,000 | 300 req/min |
| Company | $1,000/mo | 100,000 | 500 req/min |
Each endpoint has a fixed posted price per successful call — most fall between 1 and 10 credits — shown on this API's page before you run it. Exceeding the rate limit returns a 429 response. Authenticate with the X-API-Key header.