llmcapa 0.2.2__tar.gz → 0.2.4__tar.gz

{llmcapa-0.2.2 → llmcapa-0.2.4}/DEVELOP.ja.md

@@ -19,6 +19,7 @@ llmcapa/
 │   ├── models.py           # 機能データクラスと機能評価
 │   ├── registry.py         # インメモリレジストリ、ロード、およびOpenRouter取得
 │   ├── cli.py              # コマンドラインインターフェース
+│   ├── tokenizer.py        # トークンカウント（オフライン、プロバイダー別）
 │   └── data/               # 同梱されているオフライン機能データ (JSON)
 │       ├── __init__.py
 │       ├── openai.json

{llmcapa-0.2.2 → llmcapa-0.2.4}/DEVELOP.md

@@ -17,6 +17,7 @@ llmcapa/
 │   ├── models.py           # Capability dataclass and feature evaluation
 │   ├── registry.py         # In-memory registry, loading, and OpenRouter fetching
 │   ├── cli.py              # Command-line interface
+│   ├── tokenizer.py        # Token counting (offline, provider-specific)
 │   └── data/               # Bundled offline capability data (JSON)
 │       ├── __init__.py
 │       ├── openai.json

{llmcapa-0.2.2 → llmcapa-0.2.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: llmcapa
-Version: 0.2.2
+Version: 0.2.4
 Summary: Lookup capabilities (context window, modalities, features) of various LLM models offline.
 Project-URL: Homepage, https://github.com/awaku7/llmcapa
 Project-URL: Repository, https://github.com/awaku7/llmcapa
@@ -236,7 +236,7 @@ Lookup capabilities (context window, modalities, supported features) of various
 
 ## Features
 
-- **Comprehensive Bundled Data**: Offline capability data for OpenAI, Anthropic, Google (Gemini), Microsoft (Phi), Amazon (Nova/Titan), Meta (Llama), Mistral, Qwen, DeepSeek, NVIDIA, and Japanese domestic models (NTT tsuzumi, PFN PLaMo, ELYZA, etc. adopted by the Digital Agency's "GENNAI" platform).
+- **Comprehensive Bundled Data**: Offline capability data for OpenAI, Anthropic, Google (Gemini), Microsoft (Phi), Amazon (Nova/Titan), Meta (Llama), Mistral, Qwen, DeepSeek, xAI (Grok), NVIDIA, MoonshotAI (Kimi), zhipu-ai (GLM), OpenRouter, and Japanese domestic models (NTT tsuzumi, PFN PLaMo, ELYZA, SoftBank, NEC, Fujitsu, etc. adopted by the Digital Agency's "GENNAI" platform).
 - **Zero Runtime Dependencies**: Built entirely on the Python standard library.
 - **Alias Resolution**: Automatically resolves aliases and provider-specific names (e.g., `gpt-4o-2024-08-06` -> `gpt-4o`, `gemini-1.5-pro-preview-0409` -> `gemini-1.5-pro`).
 - **Advanced Feature Queries**: Check support for `vision`, `multimodal`, `chat_completion`, `responses_api`, `reasoning_effort`, `thinking_budget`, and specific input/output modalities (e.g., `image_input`, `image_output`, `audio_input`).
@@ -283,7 +283,7 @@ print(cap.supports(Feature.LLMC_FEAT_REASONING_EFFORT))   # False
 print(ReasoningEffort.LLMC_EFFORT_HIGH)                   # "high"
 # List all supported features
 print(cap.features())
-# ['chat_completion', 'function_calling', 'image', 'image_input', 'image_output', 'json_mode', 'multimodal', 'responses_api', 'streaming', 'text', 'text_input', 'text_output', 'vision']
+# ['chat_completion', 'file', 'file_input', 'function_calling', 'image', 'image_input', 'json_mode', 'multimodal', 'responses_api', 'streaming', 'text', 'text_input', 'text_output', 'vision']
 ```
 
 ### Token & Cost Estimation
@@ -317,10 +317,10 @@ gpt4o = llmcapa.get("gpt-4o")
 gpt4o_mini = llmcapa.get("gpt-4o-mini")
 gemini = llmcapa.get("gemini-3.5-flash")
 
-# gpt-4o-mini has the same context window but lacks image_output (which gpt-4o supports)
-print(gpt4o.can_be_replaced_by(gpt4o_mini))  # False
+# gpt-4o-mini has the same context window and supports all the same features
+print(gpt4o.can_be_replaced_by(gpt4o_mini))  # True
 
-# gemini-3.5-flash has a larger context window but also lacks image_output
+# gemini-3.5-flash has a larger context window but lacks responses_api (which gpt-4o supports)
 print(gpt4o.can_be_replaced_by(gemini))  # False
 
 # If we only require vision and function_calling, gemini-3.5-flash can replace gpt-4o
@@ -386,6 +386,47 @@ print(cap.context_window)  # 131072
 print(cap.pricing)         # {'input_per_1m': 0.1, 'output_per_1m': 0.32, 'currency': 'USD'}
 ```
 
+### Token Counting (Standalone)
+
+Count tokens for a single text or a list of chat messages using the best available tokenizer for a given model.
+
+```python
+# Count tokens for a text string
+import llmcapa
+tokens = llmcapa.count_tokens("Hello, world!", "gpt-4o")
+print(tokens)  # exact count if tiktoken is installed, else estimation
+
+# Count tokens for chat messages (includes overhead)
+messages = [
+    {"role": "user", "content": "Hello"},
+    {"role": "assistant", "content": "Hi there!"},
+]
+total = llmcapa.count_messages_tokens(messages, "gpt-4o")
+print(total)
+```
+
+### Programmatic Registration
+
+Register a Capability directly without a JSON file:
+
+```python
+from llmcapa import Capability
+
+cap = Capability(
+    provider="local",
+    model_id="my-model",
+    context_window=4096,
+    max_output_tokens=1024,
+    supports_function_calling=True,
+    aliases=["mm"],
+)
+llmcapa.register(cap)
+
+print(llmcapa.get("my-model").context_window)  # 4096
+```
+
+> **Note**: `llmcapa.get()` raises `ModelNotFoundError` if the model is not found.
+
 ### Custom Local Data
 
 Load your own model definitions from a local JSON file:
@@ -429,6 +470,13 @@ llmcapa list --json --no-deprecated
 # List all known providers
 llmcapa providers
 
+# Count tokens for text or messages
+llmcapa tokens gpt-4o "Hello, world!"
+llmcapa tokens gpt-4o --messages '[{"role":"user","content":"Hi"}]'
+
+# Load extra model data from a local JSON file on startup
+llmcapa --extra my_models.json show gpt-4o
+
 # Explicitly fetch and update the OpenRouter models cache (forces cache refresh)
 llmcapa update
 ```

{llmcapa-0.2.2 → llmcapa-0.2.4}/README.ja.md

@@ -4,7 +4,7 @@
 
 ## 特徴
 
-- **包括的な同梱データ**: OpenAI、Anthropic、Google (Gemini)、Microsoft (Phi)、Amazon (Nova/Titan)、Meta (Llama)、Mistral、Qwen、DeepSeek、NVIDIA、および日本の国内モデル（デジタル庁の「GENNAI」プラットフォームで採用されているNTT tsuzumi、PFN PLaMo、ELYZA、SoftBank、NEC、Fujitsuなど）のオフライン機能データを同梱しています。
+- **包括的な同梱データ**: OpenAI、Anthropic、Google (Gemini)、Microsoft (Phi)、Amazon (Nova/Titan)、Meta (Llama)、Mistral、Qwen、DeepSeek、xAI (Grok)、NVIDIA、MoonshotAI (Kimi)、zhipu-ai (GLM)、OpenRouter、および日本の国内モデル（デジタル庁の「GENNAI」プラットフォームで採用されているNTT tsuzumi、PFN PLaMo、ELYZA、SoftBank、NEC、Fujitsuなど）のオフライン機能データを同梱しています。
 - **実行時依存関係ゼロ**: Python標準ライブラリのみで動作します。外部パッケージ（`pytest` や `build` など）は開発・テスト用のみです。
 - **エイリアス解決**: エイリアスやプロバイダー固有の名前を自動的に解決します（例: `gpt-4o-2024-08-06` -> `gpt-4o`、`gemini-1.5-pro-preview-0409` -> `gemini-1.5-pro`）。
 - **高度な機能クエリ**: `vision`、`multimodal`、`chat_completion`、`responses_api`、`reasoning_effort`、`thinking_budget`、および特定の入力/出力モダリティ（例: `image_input`、`image_output`、`audio_input`）のサポート状況を確認できます。
@@ -51,7 +51,7 @@ print(cap.supports(Feature.LLMC_FEAT_REASONING_EFFORT))   # False
 print(ReasoningEffort.LLMC_EFFORT_HIGH)                   # "high"
 # サポートされているすべての機能を一覧表示
 print(cap.features())
-# ['chat_completion', 'function_calling', 'image', 'image_input', 'image_output', 'json_mode', 'multimodal', 'responses_api', 'streaming', 'text', 'text_input', 'text_output', 'vision']
+# ['chat_completion', 'file', 'file_input', 'function_calling', 'image', 'image_input', 'json_mode', 'multimodal', 'responses_api', 'streaming', 'text', 'text_input', 'text_output', 'vision']
 ```
 
 ### トークン数とコストの見積もり
@@ -85,10 +85,10 @@ gpt4o = llmcapa.get("gpt-4o")
 gpt4o_mini = llmcapa.get("gpt-4o-mini")
 gemini = llmcapa.get("gemini-3.5-flash")
 
-# gpt-4o-miniは同じコンテキストウィンドウを持ちますが、image_output（gpt-4oがサポート）がありません
-print(gpt4o.can_be_replaced_by(gpt4o_mini))  # False
+# gpt-4o-miniは同じコンテキストウィンドウを持ち、同じ機能をすべてサポートしています
+print(gpt4o.can_be_replaced_by(gpt4o_mini))  # True
 
-# gemini-3.5-flashはより大きなコンテキストウィンドウを持ちますが、同様にimage_outputがありません
+# gemini-3.5-flashはより大きなコンテキストウィンドウを持ちますが、responses_api（gpt-4oがサポート）がありません
 print(gpt4o.can_be_replaced_by(gemini))  # False
 
 # visionとfunction_callingのみが必要な場合、gemini-3.5-flashはgpt-4oを代替できます
@@ -154,6 +154,47 @@ print(cap.context_window)  # 131072
 print(cap.pricing)         # {'input_per_1m': 0.1, 'output_per_1m': 0.32, 'currency': 'USD'}
 ```
 
+### トークン数のカウント（スタンドアロン）
+
+指定されたモデルに最適なトークナイザーを使用して、単一テキストまたはチャットメッセージのトークン数をカウントします。
+
+```python
+# テキストのトークン数をカウント
+import llmcapa
+tokens = llmcapa.count_tokens("Hello, world!", "gpt-4o")
+print(tokens)  # tiktokenがインストールされていれば正確な値、それ以外は推定値
+
+# チャットメッセージのトークン数（オーバーヘッド含む）
+messages = [
+    {"role": "user", "content": "Hello"},
+    {"role": "assistant", "content": "Hi there!"},
+]
+total = llmcapa.count_messages_tokens(messages, "gpt-4o")
+print(total)
+```
+
+### プログラムによるモデル登録
+
+JSONファイルを使わずに直接Capabilityを登録:
+
+```python
+from llmcapa import Capability
+
+cap = Capability(
+    provider="local",
+    model_id="my-model",
+    context_window=4096,
+    max_output_tokens=1024,
+    supports_function_calling=True,
+    aliases=["mm"],
+)
+llmcapa.register(cap)
+
+print(llmcapa.get("my-model").context_window)  # 4096
+```
+
+> **注意**: `llmcapa.get()` はモデルが見つからない場合 `ModelNotFoundError` を送出します。
+
 ### カスタムローカルデータ
 
 ローカルのJSONファイルから独自のモデル定義をロードできます。
@@ -197,6 +238,13 @@ llmcapa list --json --no-deprecated
 # 既知のすべてのプロバイダーを一覧表示
 llmcapa providers
 
+# テキストやメッセージのトークン数をカウント
+llmcapa tokens gpt-4o "Hello, world!"
+llmcapa tokens gpt-4o --messages '[{"role":"user","content":"Hi"}]'
+
+# 起動時にローカルJSONファイルから追加モデルをロード
+llmcapa --extra my_models.json show gpt-4o
+
 # 明示的にOpenRouterモデルのキャッシュを取得・更新（キャッシュを強制リフレッシュ）
 llmcapa update
 ```

{llmcapa-0.2.2 → llmcapa-0.2.4}/README.md

@@ -4,7 +4,7 @@ Lookup capabilities (context window, modalities, supported features) of various
 
 ## Features
 
-- **Comprehensive Bundled Data**: Offline capability data for OpenAI, Anthropic, Google (Gemini), Microsoft (Phi), Amazon (Nova/Titan), Meta (Llama), Mistral, Qwen, DeepSeek, NVIDIA, and Japanese domestic models (NTT tsuzumi, PFN PLaMo, ELYZA, etc. adopted by the Digital Agency's "GENNAI" platform).
+- **Comprehensive Bundled Data**: Offline capability data for OpenAI, Anthropic, Google (Gemini), Microsoft (Phi), Amazon (Nova/Titan), Meta (Llama), Mistral, Qwen, DeepSeek, xAI (Grok), NVIDIA, MoonshotAI (Kimi), zhipu-ai (GLM), OpenRouter, and Japanese domestic models (NTT tsuzumi, PFN PLaMo, ELYZA, SoftBank, NEC, Fujitsu, etc. adopted by the Digital Agency's "GENNAI" platform).
 - **Zero Runtime Dependencies**: Built entirely on the Python standard library.
 - **Alias Resolution**: Automatically resolves aliases and provider-specific names (e.g., `gpt-4o-2024-08-06` -> `gpt-4o`, `gemini-1.5-pro-preview-0409` -> `gemini-1.5-pro`).
 - **Advanced Feature Queries**: Check support for `vision`, `multimodal`, `chat_completion`, `responses_api`, `reasoning_effort`, `thinking_budget`, and specific input/output modalities (e.g., `image_input`, `image_output`, `audio_input`).
@@ -51,7 +51,7 @@ print(cap.supports(Feature.LLMC_FEAT_REASONING_EFFORT))   # False
 print(ReasoningEffort.LLMC_EFFORT_HIGH)                   # "high"
 # List all supported features
 print(cap.features())
-# ['chat_completion', 'function_calling', 'image', 'image_input', 'image_output', 'json_mode', 'multimodal', 'responses_api', 'streaming', 'text', 'text_input', 'text_output', 'vision']
+# ['chat_completion', 'file', 'file_input', 'function_calling', 'image', 'image_input', 'json_mode', 'multimodal', 'responses_api', 'streaming', 'text', 'text_input', 'text_output', 'vision']
 ```
 
 ### Token & Cost Estimation
@@ -85,10 +85,10 @@ gpt4o = llmcapa.get("gpt-4o")
 gpt4o_mini = llmcapa.get("gpt-4o-mini")
 gemini = llmcapa.get("gemini-3.5-flash")
 
-# gpt-4o-mini has the same context window but lacks image_output (which gpt-4o supports)
-print(gpt4o.can_be_replaced_by(gpt4o_mini))  # False
+# gpt-4o-mini has the same context window and supports all the same features
+print(gpt4o.can_be_replaced_by(gpt4o_mini))  # True
 
-# gemini-3.5-flash has a larger context window but also lacks image_output
+# gemini-3.5-flash has a larger context window but lacks responses_api (which gpt-4o supports)
 print(gpt4o.can_be_replaced_by(gemini))  # False
 
 # If we only require vision and function_calling, gemini-3.5-flash can replace gpt-4o
@@ -154,6 +154,47 @@ print(cap.context_window)  # 131072
 print(cap.pricing)         # {'input_per_1m': 0.1, 'output_per_1m': 0.32, 'currency': 'USD'}
 ```
 
+### Token Counting (Standalone)
+
+Count tokens for a single text or a list of chat messages using the best available tokenizer for a given model.
+
+```python
+# Count tokens for a text string
+import llmcapa
+tokens = llmcapa.count_tokens("Hello, world!", "gpt-4o")
+print(tokens)  # exact count if tiktoken is installed, else estimation
+
+# Count tokens for chat messages (includes overhead)
+messages = [
+    {"role": "user", "content": "Hello"},
+    {"role": "assistant", "content": "Hi there!"},
+]
+total = llmcapa.count_messages_tokens(messages, "gpt-4o")
+print(total)
+```
+
+### Programmatic Registration
+
+Register a Capability directly without a JSON file:
+
+```python
+from llmcapa import Capability
+
+cap = Capability(
+    provider="local",
+    model_id="my-model",
+    context_window=4096,
+    max_output_tokens=1024,
+    supports_function_calling=True,
+    aliases=["mm"],
+)
+llmcapa.register(cap)
+
+print(llmcapa.get("my-model").context_window)  # 4096
+```
+
+> **Note**: `llmcapa.get()` raises `ModelNotFoundError` if the model is not found.
+
 ### Custom Local Data
 
 Load your own model definitions from a local JSON file:
@@ -197,6 +238,13 @@ llmcapa list --json --no-deprecated
 # List all known providers
 llmcapa providers
 
+# Count tokens for text or messages
+llmcapa tokens gpt-4o "Hello, world!"
+llmcapa tokens gpt-4o --messages '[{"role":"user","content":"Hi"}]'
+
+# Load extra model data from a local JSON file on startup
+llmcapa --extra my_models.json show gpt-4o
+
 # Explicitly fetch and update the OpenRouter models cache (forces cache refresh)
 llmcapa update
 ```

{llmcapa-0.2.2 → llmcapa-0.2.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "llmcapa"
-version = "0.2.2"
+version = "0.2.4"
 description = "Lookup capabilities (context window, modalities, features) of various LLM models offline."
 readme = "README.md"
 license = { file = "LICENSE" }

llmcapa-0.2.4/pyproject.toml.org

@@ -0,0 +1,42 @@
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[project]
+name = "llmcapa"
+version = "0.2.3"
+description = "Lookup capabilities (context window, modalities, features) of various LLM models offline."
+readme = "README.md"
+license = { file = "LICENSE" }
+requires-python = ">=3.9"
+authors = [{ name = "Hirofumi Ukawa", email = "hirofumi@ukawa.biz" }]
+keywords = ["llm", "capability", "openai", "anthropic", "gemini", "models"]
+classifiers = [
+    "Development Status :: 3 - Alpha",
+    "Intended Audience :: Developers",
+    "License :: OSI Approved :: Apache Software License",
+    "Operating System :: OS Independent",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.9",
+    "Programming Language :: Python :: 3.10",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+    "Topic :: Software Development :: Libraries",
+]
+
+[project.scripts]
+llmcapa = "llmcapa.cli:main"
+
+[project.urls]
+Homepage = "https://github.com/awaku7/llmcapa"
+Repository = "https://github.com/awaku7/llmcapa"
+"Bug Tracker" = "https://github.com/awaku7/llmcapa/issues"
+
+[project.optional-dependencies]
+openai = ["tiktoken>=0.5.0"]
+gemini = ["sentencepiece>=0.2.0"]
+all = ["tiktoken>=0.5.0", "sentencepiece>=0.2.0"]
+
+[tool.hatch.build.targets.wheel]
+packages = ["src/llmcapa"]

llmcapa-0.2.4/src/llmcapa/__init__.py

@@ -0,0 +1,92 @@
+"""llmcapa: lookup capabilities of various LLM models, fully offline.
+
+Example:
+    >>> import llmcapa
+    >>> cap = llmcapa.get("gpt-4o")
+    >>> cap.context_window
+    128000
+    >>> cap.supports("vision")
+    True
+"""
+
+from __future__ import annotations
+
+from typing import List, Optional, Union
+from pathlib import Path
+
+from .models import Capability, Feature, ReasoningEffort
+from .registry import Registry, ModelNotFoundError, default_registry
+from .tokenizer import count_tokens, count_messages_tokens
+
+__version__ = "0.2.4"
+
+__all__ = [
+    "Capability",
+    "Feature",
+    "ReasoningEffort",
+    "Registry",
+    "ModelNotFoundError",
+    "get",
+    "list_models",
+    "providers",
+    "find",
+    "search",
+    "load_extra",
+    "fetch_openrouter",
+    "register",
+    "default_registry",
+    "count_tokens",
+    "count_messages_tokens",
+    "__version__",
+]
+
+
+def get(model_id: str) -> Capability:
+    """Return the Capability for a model id or alias."""
+    return default_registry().get(model_id)
+
+
+def list_models(
+    provider: Optional[str] = None,
+    include_deprecated: bool = True,
+) -> List[Capability]:
+    """List known models, optionally filtered by provider."""
+    return default_registry().list_models(provider, include_deprecated)
+
+
+def providers() -> List[str]:
+    """Return the sorted list of known providers."""
+    return default_registry().providers()
+
+
+def find(**kwargs) -> List[Capability]:
+    """Search models by conditions. See Registry.find."""
+    return default_registry().find(**kwargs)
+
+
+def search(
+    prefix: str,
+    provider: Optional[str] = None,
+    include_deprecated: bool = False,
+    limit: Optional[int] = None,
+) -> List[Capability]:
+    """Search models by prefix matching on model_id, display_name, or aliases.
+
+    Case-insensitive prefix search. Results are sorted by (provider, model_id).
+    """
+    return default_registry().search(prefix, provider, include_deprecated, limit)
+
+
+def load_extra(path: Union[str, Path]) -> int:
+    """Load user-defined model data from a local JSON file."""
+    return default_registry().load_extra(path)
+
+
+def fetch_openrouter(cache_ttl: Optional[int] = None) -> int:
+    """Fetch all models dynamically from OpenRouter API and register them."""
+    return default_registry().fetch_openrouter(cache_ttl)
+
+
+def register(cap: Capability) -> None:
+    """Register (or override) a Capability in the default registry."""
+    default_registry().register(cap)

llmcapa-0.2.2/src/llmcapa/__init__.py → llmcapa-0.2.4/src/llmcapa/__init__.py.org

@@ -18,7 +18,7 @@ from .models import Capability, Feature, ReasoningEffort
 from .registry import Registry, ModelNotFoundError, default_registry
 from .tokenizer import count_tokens, count_messages_tokens
 
-__version__ = "0.2.0"
+__version__ = "0.2.3"
 
 __all__ = [
     "Capability",

llmcapa-0.2.4/src/llmcapa/__init__.py.org1

@@ -0,0 +1,79 @@
+"""llmcapa: lookup capabilities of various LLM models, fully offline.
+
+Example:
+    >>> import llmcapa
+    >>> cap = llmcapa.get("gpt-4o")
+    >>> cap.context_window
+    128000
+    >>> cap.supports("vision")
+    True
+"""
+
+from __future__ import annotations
+
+from typing import List, Optional, Union
+from pathlib import Path
+
+from .models import Capability, Feature, ReasoningEffort
+from .registry import Registry, ModelNotFoundError, default_registry
+from .tokenizer import count_tokens, count_messages_tokens
+
+__version__ = "0.2.3"
+
+__all__ = [
+    "Capability",
+    "Feature",
+    "ReasoningEffort",
+    "Registry",
+    "ModelNotFoundError",
+    "get",
+    "list_models",
+    "providers",
+    "find",
+    "search",
+    "load_extra",
+    "fetch_openrouter",
+    "register",
+    "default_registry",
+    "count_tokens",
+    "count_messages_tokens",
+    "__version__",
+]
+
+
+def get(model_id: str) -> Capability:
+    """Return the Capability for a model id or alias."""
+    return default_registry().get(model_id)
+
+
+def list_models(
+    provider: Optional[str] = None,
+    include_deprecated: bool = True,
+) -> List[Capability]:
+    """List known models, optionally filtered by provider."""
+    return default_registry().list_models(provider, include_deprecated)
+
+
+def providers() -> List[str]:
+    """Return the sorted list of known providers."""
+    return default_registry().providers()
+
+
+def find(**kwargs) -> List[Capability]:
+    """Search models by conditions. See Registry.find."""
+    return default_registry().find(**kwargs)
+
+
+def load_extra(path: Union[str, Path]) -> int:
+    """Load user-defined model data from a local JSON file."""
+    return default_registry().load_extra(path)
+
+
+def fetch_openrouter(cache_ttl: Optional[int] = None) -> int:
+    """Fetch all models dynamically from OpenRouter API and register them."""
+    return default_registry().fetch_openrouter(cache_ttl)
+
+
+def register(cap: Capability) -> None:
+    """Register (or override) a Capability in the default registry."""
+    default_registry().register(cap)

llmcapa-0.2.4/src/llmcapa/__init__.py.org2

@@ -0,0 +1,92 @@
+"""llmcapa: lookup capabilities of various LLM models, fully offline.
+
+Example:
+    >>> import llmcapa
+    >>> cap = llmcapa.get("gpt-4o")
+    >>> cap.context_window
+    128000
+    >>> cap.supports("vision")
+    True
+"""
+
+from __future__ import annotations
+
+from typing import List, Optional, Union
+from pathlib import Path
+
+from .models import Capability, Feature, ReasoningEffort
+from .registry import Registry, ModelNotFoundError, default_registry
+from .tokenizer import count_tokens, count_messages_tokens
+
+__version__ = "0.2.3"
+
+__all__ = [
+    "Capability",
+    "Feature",
+    "ReasoningEffort",
+    "Registry",
+    "ModelNotFoundError",
+    "get",
+    "list_models",
+    "providers",
+    "find",
+    "search",
+    "load_extra",
+    "fetch_openrouter",
+    "register",
+    "default_registry",
+    "count_tokens",
+    "count_messages_tokens",
+    "__version__",
+]
+
+
+def get(model_id: str) -> Capability:
+    """Return the Capability for a model id or alias."""
+    return default_registry().get(model_id)
+
+
+def list_models(
+    provider: Optional[str] = None,
+    include_deprecated: bool = True,
+) -> List[Capability]:
+    """List known models, optionally filtered by provider."""
+    return default_registry().list_models(provider, include_deprecated)
+
+
+def providers() -> List[str]:
+    """Return the sorted list of known providers."""
+    return default_registry().providers()
+
+
+def find(**kwargs) -> List[Capability]:
+    """Search models by conditions. See Registry.find."""
+    return default_registry().find(**kwargs)
+
+
+def search(
+    prefix: str,
+    provider: Optional[str] = None,
+    include_deprecated: bool = False,
+    limit: Optional[int] = None,
+) -> List[Capability]:
+    """Search models by prefix matching on model_id, display_name, or aliases.
+
+    Case-insensitive prefix search. Results are sorted by (provider, model_id).
+    """
+    return default_registry().search(prefix, provider, include_deprecated, limit)
+
+
+def load_extra(path: Union[str, Path]) -> int:
+    """Load user-defined model data from a local JSON file."""
+    return default_registry().load_extra(path)
+
+
+def fetch_openrouter(cache_ttl: Optional[int] = None) -> int:
+    """Fetch all models dynamically from OpenRouter API and register them."""
+    return default_registry().fetch_openrouter(cache_ttl)
+
+
+def register(cap: Capability) -> None:
+    """Register (or override) a Capability in the default registry."""
+    default_registry().register(cap)