ist_llm_model_ids.py

https://chatgpt.com/share/69498f35-4b14-8006-9a30-ff146cad3f83

Author: simonw

python/README.md

@@ -4,6 +4,26 @@ These Python scripts can be run directly from their URLs using `uv run`.
 
 Their source code is [available on GitHub](https://github.com/simonw/tools/tree/main/python).
 
+## list_llm_model_ids.py
+
+List available model IDs from OpenAI, Anthropic, and Gemini.
+
+You must have [LLM](https://llm.datasette.io/) installed and have configured the API keys for one or more of `anthorpic`, `openai`, or `gemini` using `llm keys set`.
+
+```bash
+uv run https://tools.simonwillison.net/python/list_llm_model_ids.py
+```
+Output starts:
+```
+===================
+openai (136 models)
+===================
+babbage-002
+chatgpt-4o-latest
+chatgpt-image-latest
+...
+```
+
 ## git_read_only_http.py
 
 Serve a local Git repository over HTTP in read-only mode.

python/list_llm_model_ids.py

@@ -0,0 +1,189 @@
+#!/usr/bin/env -S uv run
+# /// script
+# requires-python = ">=3.10"
+# dependencies = ["httpx"]
+# ///
+"""
+List model IDs from OpenAI, Anthropic, and Gemini using httpx.
+
+API keys are obtained via:
+  llm keys get openai
+  llm keys get anthropic
+  llm keys get gemini
+
+Providers whose key command fails are skipped.
+"""
+
+from __future__ import annotations
+
+import json
+import subprocess
+import sys
+from dataclasses import dataclass
+from typing import Iterable, Optional
+
+import httpx
+
+
+@dataclass(frozen=True)
+class ProviderResult:
+    provider: str
+    models: list[str]
+
+
+def _get_key_via_llm(provider: str) -> Optional[str]:
+    """
+    Run: llm keys get <provider>
+    Return the key string (stripped) or None if command fails.
+    """
+    try:
+        proc = subprocess.run(
+            ["llm", "keys", "get", provider],
+            check=True,
+            capture_output=True,
+            text=True,
+        )
+    except (subprocess.CalledProcessError, FileNotFoundError):
+        return None
+
+    key = (proc.stdout or "").strip()
+    return key or None
+
+
+def _print_section(title: str) -> None:
+    print("\n" + "=" * len(title))
+    print(title)
+    print("=" * len(title))
+
+
+def list_openai_models(client: httpx.Client, api_key: str) -> ProviderResult:
+    url = "https://api.openai.com/v1/models"
+    headers = {"Authorization": f"Bearer {api_key}"}
+
+    r = client.get(url, headers=headers)
+    r.raise_for_status()
+    data = r.json()
+
+    models = []
+    for m in data.get("data", []) or []:
+        mid = m.get("id")
+        if isinstance(mid, str):
+            models.append(mid)
+
+    models = sorted(set(models))
+    return ProviderResult("openai", models)
+
+
+def list_anthropic_models(client: httpx.Client, api_key: str) -> ProviderResult:
+    url = "https://api.anthropic.com/v1/models"
+    # Anthropic requires the anthropic-version header on API requests.
+    headers = {
+        "x-api-key": api_key,
+        "anthropic-version": "2023-06-01",
+    }
+
+    models: list[str] = []
+    after_id: Optional[str] = None
+
+    while True:
+        params = {}
+        if after_id:
+            params["after_id"] = after_id
+
+        r = client.get(url, headers=headers, params=params)
+        r.raise_for_status()
+        payload = r.json()
+
+        for m in payload.get("data", []) or []:
+            mid = m.get("id")
+            if isinstance(mid, str):
+                models.append(mid)
+
+        has_more = bool(payload.get("has_more"))
+        after_id = payload.get("last_id") if has_more else None
+        if not has_more or not isinstance(after_id, str) or not after_id:
+            break
+
+    models = sorted(set(models))
+    return ProviderResult("anthropic", models)
+
+
+def list_gemini_models(client: httpx.Client, api_key: str) -> ProviderResult:
+    url = "https://generativelanguage.googleapis.com/v1beta/models"
+
+    models: list[str] = []
+    page_token: Optional[str] = None
+
+    while True:
+        params = {"key": api_key}
+        if page_token:
+            params["pageToken"] = page_token
+
+        r = client.get(url, params=params)
+        r.raise_for_status()
+        payload = r.json()
+
+        for m in payload.get("models", []) or []:
+            # Gemini uses "name" like "models/gemini-2.0-flash"
+            name = m.get("name")
+            if isinstance(name, str):
+                models.append(name)
+
+        page_token = payload.get("nextPageToken")
+        if not isinstance(page_token, str) or not page_token:
+            break
+
+    models = sorted(set(models))
+    return ProviderResult("gemini", models)
+
+
+def main(argv: list[str]) -> int:
+    # Optional: --json to emit machine-readable output.
+    as_json = "--json" in argv
+
+    # Try to get keys. Skip providers that error.
+    keys = {
+        "openai": _get_key_via_llm("openai"),
+        "anthropic": _get_key_via_llm("anthropic"),
+        "gemini": _get_key_via_llm("gemini"),
+    }
+
+    available = {k: v for k, v in keys.items() if v}
+    if not available:
+        print("No API keys found via `llm keys get {openai,anthropic,gemini}`. Nothing to do.", file=sys.stderr)
+        return 2
+
+    results: list[ProviderResult] = []
+
+    with httpx.Client(timeout=httpx.Timeout(30.0)) as client:
+        for provider, key in available.items():
+            try:
+                if provider == "openai":
+                    results.append(list_openai_models(client, key))
+                elif provider == "anthropic":
+                    results.append(list_anthropic_models(client, key))
+                elif provider == "gemini":
+                    results.append(list_gemini_models(client, key))
+            except httpx.HTTPError as e:
+                # Don't kill the whole run if one provider fails.
+                print(f"[{provider}] HTTP error: {e}", file=sys.stderr)
+            except (ValueError, json.JSONDecodeError) as e:
+                print(f"[{provider}] JSON parse error: {e}", file=sys.stderr)
+
+    if as_json:
+        out = {r.provider: r.models for r in results}
+        print(json.dumps(out, indent=2, sort_keys=True))
+        return 0
+
+    # Pretty output
+    for r in results:
+        _print_section(f"{r.provider} ({len(r.models)} models)")
+        for mid in r.models:
+            print(mid)
+
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main(sys.argv[1:]))
+