### Install Dependencies and Set API Keys

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

Installs project dependencies and sets required environment variables for API keys. Ensure all necessary keys are exported before running the script.

```bash
pip install -r src/requirements.txt

export GROQ_API_KEY="gsk_..."
export CLOUDFLARE_ACCOUNT_ID="abc123"
export CLOUDFLARE_API_KEY="cfkey..."
export HYPERBOLIC_API_KEY="hk-..."
export GCP_PROJECT_ID="my-project"
export GOOGLE_APPLICATION_CREDENTIALS="/path/to/adc.json"
export LAMBDA_API_KEY="lk-..."
export SCALEWAY_API_KEY="sk-..."
export COHERE_API_KEY="co-..."
```

--------------------------------

### GitHub Actions Workflow for Daily README Update

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

A GitHub Actions workflow that automatically updates the README.md file daily. It checks out the code, sets up Python, installs dependencies, authenticates to GCP, runs the generation script with injected secrets, and opens a pull request.

```yaml
# .github/workflows/update-readme.yml (key excerpt)
on:
  schedule:
    - cron: "0 0 * * *"   # runs every day at midnight UTC
  workflow_dispatch:        # also triggerable manually from the Actions UI

jobs:
  update-readme:
    steps:
      - uses: actions/checkout@v6
      - uses: actions/setup-python@v6
        with:
          python-version: "3.12"
      - run: pip install -r src/requirements.txt
      - uses: google-github-actions/auth@v3  # Workload Identity — no long-lived key
        with:
          workload_identity_provider: "projects/.../providers/cheahjs-org"
          project_id: ${{ secrets.GCP_PROJECT }}
      - run: python -u src/pull_available_models.py
        env:
          GROQ_API_KEY:          ${{ secrets.GROQ_API_KEY }}
          CLOUDFLARE_ACCOUNT_ID: ${{ secrets.CLOUDFLARE_ACCOUNT_ID }}
          CLOUDFLARE_API_KEY:    ${{ secrets.CLOUDFLARE_API_KEY }}
          HYPERBOLIC_API_KEY:    ${{ secrets.HYPERBOLIC_API_KEY }}
          GCP_PROJECT_ID:        ${{ secrets.GCP_PROJECT_ID }}
          LAMBDA_API_KEY:        ${{ secrets.LAMBDA_API_KEY }}
          SCALEWAY_API_KEY:      ${{ secrets.SCALEWAY_API_KEY }}
          COHERE_API_KEY:        ${{ secrets.COHERE_API_KEY }}
      - uses: peter-evans/create-pull-request@v8
        with:
          title: "Update README with latest models"
          branch: update-readme
          base: main
```

--------------------------------

### Run README Generation Script

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

Executes the main script to generate the README.md file. Can be run sequentially or in parallel for faster execution. Logs any missing model IDs from MODEL_TO_NAME_MAPPING.

```bash
# Run sequentially (default)
python src/pull_available_models.py

# Run provider fetches in parallel (faster)
FETCH_CONCURRENTLY=true python src/pull_available_models.py
```

--------------------------------

### Fetch Google AI Studio Quotas

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

Fetches free-tier token and request quotas for Gemini/Gemma models using the google-cloud-quotas SDK. Requires Application Default Credentials (ADC) or GOOGLE_APPLICATION_CREDENTIALS.

```python
import os
from pull_available_models import fetch_gemini_limits, create_logger

# Requires Application Default Credentials (ADC) or GOOGLE_APPLICATION_CREDENTIALS
os.environ["GCP_PROJECT_ID"] = "my-gcp-project"

logger = create_logger("Google AI Studio")
limits = fetch_gemini_limits(logger)

# Example output:
# {
#   "gemini-2.5-flash": {"tokens/minute": 250000, "requests/minute": 10},
#   "gemma-3-27b":      {"tokens/minute": 15000,  "requests/minute": 2},
#   ...
# }
print(limits.get("gemini-2.5-flash"))
```

--------------------------------

### Fetch Free OpenRouter Models

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

The `fetch_openrouter_models` function retrieves all models from the OpenRouter API, filtering for those that are free (completion and prompt pricing are zero) and have a ':free' suffix in their ID. It also excludes models listed in `OPENROUTER_IGNORED_MODELS`.

```python
from pull_available_models import fetch_openrouter_models, create_logger

logger = create_logger("OpenRouter")
models = fetch_openrouter_models(logger)

# Each model: {"id": "deepseek/deepseek-r1:free", "name": "DeepSeek R1",
#              "limits": {"requests/minute": 20, "requests/day": 50}}
for m in models:
    print(f"  - {m['name']} ({m['id']})")
```

--------------------------------

### Fetch Groq Models with Rate Limits

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

The `fetch_groq_models` function queries the Groq API for available models and their rate limits by inspecting response headers from inference calls. It also handles Whisper models separately.

```python
import os, logging, requests
from pull_available_models import fetch_groq_models, create_logger

os.environ["GROQ_API_KEY"] = "gsk_..."

logger = create_logger("Groq")
models = fetch_groq_models(logger)

# Example output structure:
# [
#   {"id": "llama-3.3-70b-versatile", "name": "Llama 3.3 70B", "limits": {"requests/day": 1000, "tokens/minute": 6000}},
#   {"id": "whisper-large-v3",         "name": "Whisper Large v3", "limits": {"audio-seconds/minute": 7200, "requests/day": 2000}},
#   ...
# ]
for m in models[:3]:
    print(f"{m['name']:40s}  {m['limits']}")
```

--------------------------------

### Fetch Hyperbolic Inference Models

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

Retrieves models from the Hyperbolic /v1/models endpoint, filtering out non-LLM entries. All remaining models share a flat limit of 60 requests per minute. Requires HYPERBOLIC_API_KEY.

```python
import os
from pull_available_models import fetch_hyperbolic_models, create_logger

os.environ["HYPERBOLIC_API_KEY"] = "hk-..."

logger = create_logger("Hyperbolic")
models = fetch_hyperbolic_models(logger)

# [{"id": "meta-llama/Llama-3.3-70B-Instruct", "name": "Llama 3.3 70B Instruct",
#   "limits": {"requests/minute": 60}}, ...]

```

--------------------------------

### Map LLM Model IDs to Human-Readable Names

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

Use the MODEL_TO_NAME_MAPPING dictionary to convert raw model IDs into display names. It supports various provider prefixes and suffixes. A helper function `get_model_name` provides a fallback to the raw ID for unknown models.

```python
from data import MODEL_TO_NAME_MAPPING

# Look up a Cloudflare-prefixed model ID
print(MODEL_TO_NAME_MAPPING["@cf/meta/llama-3.3-70b-instruct-fp8-fast"])
# Output: "Llama 3.3 70B Instruct (FP8)"

# Look up an OpenRouter free-tier model
print(MODEL_TO_NAME_MAPPING["deepseek/deepseek-r1:free"])
# Output: "DeepSeek R1"

# Graceful fallback for unknown IDs (handled by get_model_name())
def get_model_name(model_id: str) -> str:
    model_id = model_id.lower()
    return MODEL_TO_NAME_MAPPING.get(model_id, model_id)  # falls back to raw ID

print(get_model_name("some/unknown-model-id"))
# Output: "some/unknown-model-id"
```

--------------------------------

### Generate GitHub-style Markdown TOC

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

Scans Markdown content for '##' and '###' headings to create a nested bullet-list TOC with GitHub-compatible anchors.

```python
from pull_available_models import generate_toc

sample_md = """
# Free LLM API Resources

## Free Providers

### OpenRouter

### Groq

## Providers with Trial Credits

### Hyperbolic
"""

toc = generate_toc(sample_md)
print(toc)
# Output:
# - [Free Providers](#free-providers)
#   - [OpenRouter](#openrouter)
#   - [Groq](#groq)
# - [Providers with Trial Credits](#providers-with-trial-credits)

```

--------------------------------

### Scrape GitHub Marketplace Models

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

Collects all models from the GitHub Marketplace by paginating through the API. Maps raw 'name' to 'friendly_name' directly from the GitHub response.

```python
from pull_available_models import fetch_github_models, create_logger

logger = create_logger("GitHub")
models = fetch_github_models(logger)

# [{"id": "openai-gpt-4o", "name": "GPT-4o"}, {"id": "meta-llama-3-70b-instruct", "name": "Meta Llama 3 70B Instruct"}, ...]
print(f"Total GitHub models: {len(models)}")
for m in models[:5]:
    print(f"  {m['name']}")
```

--------------------------------

### Format Rate Limits for Markdown

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

Converts a model's limits dictionary into a human-readable string for Markdown display. The separator defaults to '<br>' for HTML tables but can be overridden.

```python
from pull_available_models import get_human_limits

model = {
    "id": "llama-3.3-70b-versatile",
    "name": "Llama 3.3 70B",
    "limits": {"requests/day": 1000, "tokens/minute": 6000},
}

html_limits = get_human_limits(model)
print(html_limits)
# Output: "1,000 requests/day<br>6,000 tokens/minute"

plain_limits = get_human_limits(model, seperator="\n")
print(plain_limits)
# Output:
# 1,000 requests/day
# 6,000 tokens/minute

# Model without limits returns empty string
print(get_human_limits({"id": "x", "name": "X"}))
# Output: ""
```

--------------------------------

### Fetch Cohere Chat Models

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

Paginates through the Cohere /v1/models endpoint to fetch active chat models, skipping deprecated ones. Requires COHERE_API_KEY.

```python
import os
from pull_available_models import fetch_cohere_models, create_logger

os.environ["COHERE_API_KEY"] = "co-..."

logger = create_logger("Cohere")
models = fetch_cohere_models(logger)

# [{"id": "command-r-plus", "name": "Command R+"}, {"id": "command-r", "name": "Command R"}, ...]
print(f"Active Cohere chat models: {len(models)}")
```

--------------------------------

### Fetch Cloudflare Workers AI Models

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

Queries the Cloudflare Account AI model search endpoint for text-generation models. Requires CLOUDFLARE_ACCOUNT_ID and CLOUDFLARE_API_KEY environment variables. Cloudflare enforces a global limit of 10,000 neurons/day.

```python
import os
from pull_available_models import fetch_cloudflare_models, create_logger

os.environ["CLOUDFLARE_ACCOUNT_ID"] = "abc123"
os.environ["CLOUDFLARE_API_KEY"]    = "cfkey..."

logger = create_logger("Cloudflare")
models = fetch_cloudflare_models(logger)

# Models have no "limits" key — Cloudflare enforces 10,000 neurons/day globally
# [{"id": "@cf/meta/llama-3.3-70b-instruct-fp8-fast", "name": "Llama 3.3 70B Instruct (FP8)"}, ...]
for m in models:
    print(m["name"])
```

--------------------------------

### Manage Ignored LLM Models Per Provider

Source: https://context7.com/cheahjs/free-llm-api-resources/llms.txt

Utilize sets like HYPERBOLIC_IGNORED_MODELS, LAMBDA_IGNORED_MODELS, and OPENROUTER_IGNORED_MODELS to exclude specific model IDs from README generation. This is useful for filtering out non-text-generation, deprecated, or unusable models.

```python
from data import HYPERBOLIC_IGNORED_MODELS, LAMBDA_IGNORED_MODELS, OPENROUTER_IGNORED_MODELS

# Check whether a model should be skipped for Hyperbolic
model_id = "FLUX.1-dev"
if model_id in HYPERBOLIC_IGNORED_MODELS:
    print(f"Skipping {model_id} — not a text-generation model")
# Output: Skipping FLUX.1-dev — not a text-generation model

# Check OpenRouter ignore list (all are unusable Gemini experimental free models)
print("google/gemini-exp-1206:free" in OPENROUTER_IGNORED_MODELS)  # True
print("deepseek/deepseek-r1:free" in OPENROUTER_IGNORED_MODELS)    # False

# Lambda ignore list targets specific broken model slugs
print(LAMBDA_IGNORED_MODELS)
# Output: {'lfm-40b-vllm', 'hermes3-405b-fp8-128k'}
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.