Skip to content

client_cache

langroid/language_models/client_cache.py

Client caching/singleton pattern for LLM clients to prevent connection pool exhaustion.

get_openai_client(api_key, base_url=None, organization=None, timeout=120.0, default_headers=None)

Get or create a singleton OpenAI client with the given configuration.

Parameters:

Name Type Description Default
api_key str

OpenAI API key

 required
base_url Optional[str]

Optional base URL for API

 None
organization Optional[str]

Optional organization ID

 None
timeout Union[float, Timeout]

Request timeout

 120.0
default_headers Optional[Dict[str, str]]

Optional default headers

 None

Returns:

Type Description
OpenAI

OpenAI client instance
Source code in langroid/language_models/client_cache.py 
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
def get_openai_client(
    api_key: str,
    base_url: Optional[str] = None,
    organization: Optional[str] = None,
    timeout: Union[float, Timeout] = 120.0,
    default_headers: Optional[Dict[str, str]] = None,
) -> OpenAI:
    """
    Get or create a singleton OpenAI client with the given configuration.

    Args:
        api_key: OpenAI API key
        base_url: Optional base URL for API
        organization: Optional organization ID
        timeout: Request timeout
        default_headers: Optional default headers

    Returns:
        OpenAI client instance
    """
    if isinstance(timeout, (int, float)):
        timeout = Timeout(timeout)

    cache_key = _get_cache_key(
        "openai",
        api_key=api_key,
        base_url=base_url,
        organization=organization,
        timeout=timeout,
        default_headers=default_headers,
    )

    if cache_key in _client_cache:
        return cast(OpenAI, _client_cache[cache_key])

    client = OpenAI(
        api_key=api_key,
        base_url=base_url,
        organization=organization,
        timeout=timeout,
        default_headers=default_headers,
    )

    _client_cache[cache_key] = client
    _all_clients.add(client)
    return client

get_async_openai_client(api_key, base_url=None, organization=None, timeout=120.0, default_headers=None)

Get or create a singleton AsyncOpenAI client with the given configuration.

Parameters:

Name Type Description Default
api_key str

OpenAI API key

 required
base_url Optional[str]

Optional base URL for API

 None
organization Optional[str]

Optional organization ID

 None
timeout Union[float, Timeout]

Request timeout

 120.0
default_headers Optional[Dict[str, str]]

Optional default headers

 None

Returns:

Type Description
AsyncOpenAI

AsyncOpenAI client instance
Source code in langroid/language_models/client_cache.py 
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
def get_async_openai_client(
    api_key: str,
    base_url: Optional[str] = None,
    organization: Optional[str] = None,
    timeout: Union[float, Timeout] = 120.0,
    default_headers: Optional[Dict[str, str]] = None,
) -> AsyncOpenAI:
    """
    Get or create a singleton AsyncOpenAI client with the given configuration.

    Args:
        api_key: OpenAI API key
        base_url: Optional base URL for API
        organization: Optional organization ID
        timeout: Request timeout
        default_headers: Optional default headers

    Returns:
        AsyncOpenAI client instance
    """
    if isinstance(timeout, (int, float)):
        timeout = Timeout(timeout)

    cache_key = _get_cache_key(
        "async_openai",
        api_key=api_key,
        base_url=base_url,
        organization=organization,
        timeout=timeout,
        default_headers=default_headers,
    )

    if cache_key in _client_cache:
        return cast(AsyncOpenAI, _client_cache[cache_key])

    client = AsyncOpenAI(
        api_key=api_key,
        base_url=base_url,
        organization=organization,
        timeout=timeout,
        default_headers=default_headers,
    )

    _client_cache[cache_key] = client
    _all_clients.add(client)
    return client

get_groq_client(api_key)

Get or create a singleton Groq client with the given configuration.

Parameters:

Name Type Description Default
api_key str

Groq API key

 required

Returns:

Type Description
Groq

Groq client instance
Source code in langroid/language_models/client_cache.py 
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
def get_groq_client(api_key: str) -> Groq:
    """
    Get or create a singleton Groq client with the given configuration.

    Args:
        api_key: Groq API key

    Returns:
        Groq client instance
    """
    cache_key = _get_cache_key("groq", api_key=api_key)

    if cache_key in _client_cache:
        return cast(Groq, _client_cache[cache_key])

    client = Groq(api_key=api_key)
    _client_cache[cache_key] = client
    _all_clients.add(client)
    return client

get_async_groq_client(api_key)

Get or create a singleton AsyncGroq client with the given configuration.

Parameters:

Name Type Description Default
api_key str

Groq API key

 required

Returns:

Type Description
AsyncGroq

AsyncGroq client instance
Source code in langroid/language_models/client_cache.py 
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
def get_async_groq_client(api_key: str) -> AsyncGroq:
    """
    Get or create a singleton AsyncGroq client with the given configuration.

    Args:
        api_key: Groq API key

    Returns:
        AsyncGroq client instance
    """
    cache_key = _get_cache_key("async_groq", api_key=api_key)

    if cache_key in _client_cache:
        return cast(AsyncGroq, _client_cache[cache_key])

    client = AsyncGroq(api_key=api_key)
    _client_cache[cache_key] = client
    _all_clients.add(client)
    return client

get_cerebras_client(api_key)

Get or create a singleton Cerebras client with the given configuration.

Parameters:

Name Type Description Default
api_key str

Cerebras API key

 required

Returns:

Type Description
Cerebras

Cerebras client instance
Source code in langroid/language_models/client_cache.py 
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
def get_cerebras_client(api_key: str) -> Cerebras:
    """
    Get or create a singleton Cerebras client with the given configuration.

    Args:
        api_key: Cerebras API key

    Returns:
        Cerebras client instance
    """
    cache_key = _get_cache_key("cerebras", api_key=api_key)

    if cache_key in _client_cache:
        return cast(Cerebras, _client_cache[cache_key])

    client = Cerebras(api_key=api_key)
    _client_cache[cache_key] = client
    _all_clients.add(client)
    return client

get_async_cerebras_client(api_key)

Get or create a singleton AsyncCerebras client with the given configuration.

Parameters:

Name Type Description Default
api_key str

Cerebras API key

 required

Returns:

Type Description
AsyncCerebras

AsyncCerebras client instance
Source code in langroid/language_models/client_cache.py 
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
def get_async_cerebras_client(api_key: str) -> AsyncCerebras:
    """
    Get or create a singleton AsyncCerebras client with the given configuration.

    Args:
        api_key: Cerebras API key

    Returns:
        AsyncCerebras client instance
    """
    cache_key = _get_cache_key("async_cerebras", api_key=api_key)

    if cache_key in _client_cache:
        return cast(AsyncCerebras, _client_cache[cache_key])

    client = AsyncCerebras(api_key=api_key)
    _client_cache[cache_key] = client
    _all_clients.add(client)
    return client