Merge 90d8171a65 into 482f3c37ed

litellm/constants.py

@@ -1095,6 +1095,7 @@ openai_compatible_endpoints: Final[list[str]] = [
     "https://gigachat.devices.sberbank.ru/api/v1",
     "https://api.openinfer.ai/v1",
     "https://vanchin.streamlake.ai/api/gateway/v1/endpoints",
+    "https://api.mallardai.co.uk/v1",
 ]
 
 
@@ -1179,6 +1180,7 @@ openai_compatible_providers: Final[list[str]] = [
     "sail",
     "reka",
     "streamlake",
+    "mallard",
 ]
 
 OPENAI_AUDIO_TRANSCRIPTION_PROVIDERS: Final = frozenset({"openai"} | frozenset(openai_compatible_providers))

litellm/llms/openai_like/providers.json

@@ -302,5 +302,10 @@
       "preserve_cache_control": true
     },
     "supported_endpoints": ["/v1/chat/completions"]
+  },
+  "mallard": {
+    "base_url": "https://api.mallardai.co.uk/v1",
+    "api_key_env": "MALLARD_API_KEY",
+    "api_base_env": "MALLARD_API_BASE"
   }
 }

litellm/model_prices_and_context_window_backup.json

@@ -77245,6 +77245,23 @@
         "supports_reasoning": true,
         "source": "https://api.openinfer.ai/v1/models"
     },
+    "mallard/qwen/qwen3.6-35b-a3b": {
+        "max_tokens": 235929,
+        "max_input_tokens": 262144,
+        "max_output_tokens": 235929,
+        "input_cost_per_token": 1.06944e-07,
+        "cache_read_input_token_cost": 8.0208e-08,
+        "output_cost_per_token": 1.0026e-06,
+        "litellm_provider": "mallard",
+        "mode": "chat",
+        "supports_function_calling": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_prompt_caching": true,
+        "source": "https://mallardai.co.uk/pricing/"
+    },
     "openinfer/@oi/Qwen3.5-27B": {
         "max_input_tokens": 32768,
         "max_output_tokens": 32768,

litellm/provider_endpoints_support_backup.json

@@ -5276,6 +5276,24 @@
         "generateContent": false
       }
     },
+    "mallard": {
+      "display_name": "Mallard AI (`mallard`)",
+      "url": "https://docs.litellm.ai/docs/providers/mallard",
+      "endpoints": {
+        "chat_completions": true,
+        "messages": true,
+        "responses": true,
+        "embeddings": false,
+        "image_generations": false,
+        "audio_transcriptions": false,
+        "audio_speech": false,
+        "moderations": false,
+        "batches": false,
+        "rerank": false,
+        "a2a": false,
+        "interactions": false
+      }
+    },
     "openinfer": {
       "display_name": "OpenInfer (`openinfer`)",
       "url": "https://docs.litellm.ai/docs/providers/openinfer",

litellm/proxy/public_endpoints/provider_create_fields.json

@@ -3539,6 +3539,34 @@
     ],
     "default_model_placeholder": "gpt-3.5-turbo"
   },
+  {
+    "provider": "MALLARD",
+    "provider_display_name": "Mallard AI",
+    "litellm_provider": "mallard",
+    "credential_fields": [
+      {
+        "key": "api_base",
+        "label": "API Base",
+        "placeholder": "https://api.mallardai.co.uk/v1",
+        "tooltip": null,
+        "required": false,
+        "field_type": "text",
+        "options": null,
+        "default_value": null
+      },
+      {
+        "key": "api_key",
+        "label": "API Key",
+        "placeholder": null,
+        "tooltip": null,
+        "required": true,
+        "field_type": "password",
+        "options": null,
+        "default_value": null
+      }
+    ],
+    "default_model_placeholder": "mallard/qwen/qwen3.6-35b-a3b"
+  },
   {
     "provider": "REKA",
     "provider_display_name": "Reka",

litellm/types/utils.py

@@ -4756,6 +4756,7 @@ class LlmProviders(str, Enum):
     SAIL = "sail"
     REKA = "reka"
     STREAMLAKE = "streamlake"
+    MALLARD = "mallard"
     LITELLM_AGENT = "litellm_agent"
     CURSOR = "cursor"
     BEDROCK_MANTLE = "bedrock_mantle"

model_prices_and_context_window.json

@@ -77245,6 +77245,23 @@
         "supports_reasoning": true,
         "source": "https://api.openinfer.ai/v1/models"
     },
+    "mallard/qwen/qwen3.6-35b-a3b": {
+        "max_tokens": 235929,
+        "max_input_tokens": 262144,
+        "max_output_tokens": 235929,
+        "input_cost_per_token": 1.06944e-07,
+        "cache_read_input_token_cost": 8.0208e-08,
+        "output_cost_per_token": 1.0026e-06,
+        "litellm_provider": "mallard",
+        "mode": "chat",
+        "supports_function_calling": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_prompt_caching": true,
+        "source": "https://mallardai.co.uk/pricing/"
+    },
     "openinfer/@oi/Qwen3.5-27B": {
         "max_input_tokens": 32768,
         "max_output_tokens": 32768,

provider_endpoints_support.json

@@ -5276,6 +5276,24 @@
         "generateContent": false
       }
     },
+    "mallard": {
+      "display_name": "Mallard AI (`mallard`)",
+      "url": "https://docs.litellm.ai/docs/providers/mallard",
+      "endpoints": {
+        "chat_completions": true,
+        "messages": true,
+        "responses": true,
+        "embeddings": false,
+        "image_generations": false,
+        "audio_transcriptions": false,
+        "audio_speech": false,
+        "moderations": false,
+        "batches": false,
+        "rerank": false,
+        "a2a": false,
+        "interactions": false
+      }
+    },
     "openinfer": {
       "display_name": "OpenInfer (`openinfer`)",
       "url": "https://docs.litellm.ai/docs/providers/openinfer",

tests/unit/llms/openai_like/test_mallard_provider.py

@@ -0,0 +1,211 @@
+import json
+from pathlib import Path
+from typing import Final
+
+import pytest
+import respx
+
+import litellm
+from litellm.caching.llm_caching_handler import LLMClientCache
+
+_CHAT_COMPLETION: Final = {
+    "id": "chatcmpl_mallard",
+    "object": "chat.completion",
+    "created": 1_790_000_000,
+    "model": "qwen/qwen3.6-35b-a3b",
+    "choices": [
+        {
+            "index": 0,
+            "message": {"role": "assistant", "content": "Hello from Mallard AI"},
+            "finish_reason": "stop",
+        }
+    ],
+    "usage": {"prompt_tokens": 4, "completion_tokens": 3, "total_tokens": 7},
+}
+
+
+def test_mallard_is_a_registered_provider():
+    assert litellm.LlmProviders.MALLARD.value == "mallard"
+    assert "mallard" in litellm.provider_list
+    assert "mallard" in litellm.constants.openai_compatible_providers
+
+
+def test_mallard_provider_resolution(monkeypatch: pytest.MonkeyPatch):
+    from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider
+
+    monkeypatch.setenv("MALLARD_API_KEY", "mallard-test-key")
+
+    model, provider, api_key, api_base = get_llm_provider(
+        model="mallard/qwen/qwen3.6-35b-a3b",
+        custom_llm_provider=None,
+        api_base=None,
+        api_key=None,
+    )
+
+    assert model == "qwen/qwen3.6-35b-a3b"
+    assert provider == "mallard"
+    assert api_key == "mallard-test-key"
+    assert api_base == "https://api.mallardai.co.uk/v1"
+
+
+def test_mallard_provider_keeps_explicit_credentials(monkeypatch: pytest.MonkeyPatch):
+    from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider
+
+    monkeypatch.setenv("MALLARD_API_KEY", "mallard-env-key")
+    monkeypatch.setenv("MALLARD_API_BASE", "https://mallard.env.example/v1")
+
+    _, provider, api_key, api_base = get_llm_provider(
+        model="mallard/qwen/qwen3.6-35b-a3b",
+        custom_llm_provider=None,
+        api_base="https://mallard.internal.example/v1",
+        api_key="mallard-explicit-key",
+    )
+
+    assert provider == "mallard"
+    assert api_key == "mallard-explicit-key"
+    assert api_base == "https://mallard.internal.example/v1"
+
+
+def test_mallard_api_base_env_overrides_default(monkeypatch: pytest.MonkeyPatch):
+    from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider
+
+    monkeypatch.setenv("MALLARD_API_KEY", "mallard-env-key")
+    monkeypatch.setenv("MALLARD_API_BASE", "https://mallard.env.example/v1")
+
+    _, provider, _, api_base = get_llm_provider(
+        model="mallard/qwen/qwen3.6-35b-a3b",
+        custom_llm_provider=None,
+        api_base=None,
+        api_key=None,
+    )
+
+    assert provider == "mallard"
+    assert api_base == "https://mallard.env.example/v1"
+
+
+def test_mallard_api_base_autodetects_provider(monkeypatch: pytest.MonkeyPatch):
+    from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider
+
+    monkeypatch.setenv("MALLARD_API_KEY", "mallard-env-key")
+
+    model, provider, api_key, api_base = get_llm_provider(
+        model="qwen/qwen3.6-35b-a3b",
+        custom_llm_provider=None,
+        api_base="https://api.mallardai.co.uk/v1",
+        api_key=None,
+    )
+
+    assert model == "qwen/qwen3.6-35b-a3b"
+    assert provider == "mallard"
+    assert api_key == "mallard-env-key"
+    assert api_base == "https://api.mallardai.co.uk/v1"
+
+
+def test_mallard_is_available_in_add_model_form():
+    fields_path = Path(litellm.__file__).parent / "proxy" / "public_endpoints" / "provider_create_fields.json"
+    providers = json.loads(fields_path.read_text())
+    mallard = next(provider for provider in providers if provider["litellm_provider"] == "mallard")
+
+    assert mallard["provider"] == "MALLARD"
+    assert mallard["provider_display_name"] == "Mallard AI"
+    assert mallard["default_model_placeholder"] == "mallard/qwen/qwen3.6-35b-a3b"
+    assert {field["key"]: field["required"] for field in mallard["credential_fields"]} == {
+        "api_base": False,
+        "api_key": True,
+    }
+
+
+def test_mallard_supported_endpoints():
+    expected: Final = {
+        "chat_completions": True,
+        "messages": True,
+        "responses": True,
+        "embeddings": False,
+        "image_generations": False,
+        "audio_transcriptions": False,
+        "audio_speech": False,
+        "moderations": False,
+        "batches": False,
+        "rerank": False,
+        "a2a": False,
+        "interactions": False,
+    }
+    backup_path = Path(litellm.__file__).parent / "provider_endpoints_support_backup.json"
+    root_path = Path(litellm.__file__).parent.parent / "provider_endpoints_support.json"
+
+    assert json.loads(backup_path.read_text())["providers"]["mallard"]["endpoints"] == expected
+    assert json.loads(root_path.read_text())["providers"]["mallard"]["endpoints"] == expected
+
+
+def test_mallard_chat_completion_request():
+    with respx.mock() as upstream:
+        route: Final = upstream.post("https://api.mallardai.co.uk/v1/chat/completions").respond(200, json=_CHAT_COMPLETION)
+        response: Final = litellm.completion(
+            model="mallard/qwen/qwen3.6-35b-a3b",
+            messages=[{"role": "user", "content": "Say hello"}],
+            api_key="mallard-test-key",
+        )
+
+    request: Final = route.calls.last.request
+    body: Final = json.loads(request.content)
+    assert route.call_count == 1
+    assert str(request.url) == "https://api.mallardai.co.uk/v1/chat/completions"
+    assert request.headers["authorization"] == "Bearer mallard-test-key"
+    assert body["model"] == "qwen/qwen3.6-35b-a3b"
+    assert body["messages"] == [{"role": "user", "content": "Say hello"}]
+    assert response.choices[0].message.content == "Hello from Mallard AI"
+
+
+def test_mallard_responses_request_is_bridged_to_chat_completions():
+    with respx.mock() as upstream:
+        route: Final = upstream.post("https://api.mallardai.co.uk/v1/chat/completions").respond(200, json=_CHAT_COMPLETION)
+        response: Final = litellm.responses(
+            model="mallard/qwen/qwen3.6-35b-a3b",
+            input="Say hello",
+            api_key="mallard-test-key",
+        )
+
+    request: Final = route.calls.last.request
+    body: Final = json.loads(request.content)
+    assert route.call_count == 1
+    assert request.headers["authorization"] == "Bearer mallard-test-key"
+    assert body["model"] == "qwen/qwen3.6-35b-a3b"
+    assert body["messages"] == [{"role": "user", "content": "Say hello"}]
+    assert response.output[0].content[0].text == "Hello from Mallard AI"
+
+
+@pytest.mark.asyncio
+async def test_mallard_anthropic_messages_request_is_bridged_to_chat_completions(monkeypatch: pytest.MonkeyPatch):
+    monkeypatch.setattr(litellm, "disable_aiohttp_transport", True)
+    monkeypatch.setattr(litellm, "in_memory_llm_clients_cache", LLMClientCache())
+    with respx.mock() as upstream:
+        route: Final = upstream.post("https://api.mallardai.co.uk/v1/chat/completions").respond(200, json=_CHAT_COMPLETION)
+        response: Final = await litellm.anthropic.messages.acreate(
+            model="mallard/qwen/qwen3.6-35b-a3b",
+            messages=[{"role": "user", "content": "Say hello"}],
+            max_tokens=32,
+            api_key="mallard-test-key",
+        )
+
+    request: Final = route.calls.last.request
+    body: Final = json.loads(request.content)
+    assert route.call_count == 1
+    assert request.headers["authorization"] == "Bearer mallard-test-key"
+    assert body["model"] == "qwen/qwen3.6-35b-a3b"
+    assert body["messages"] == [{"role": "user", "content": "Say hello"}]
+    assert body["max_tokens"] == 32
+    assert response["content"][0]["text"] == "Hello from Mallard AI"
+
+
+def test_mallard_model_prices_and_limits():
+    info: Final = litellm.get_model_info("mallard/qwen/qwen3.6-35b-a3b")
+
+    assert info["litellm_provider"] == "mallard"
+    assert info["max_input_tokens"] == 262_144
+    assert info["max_output_tokens"] == 235_929
+    assert info["input_cost_per_token"] == pytest.approx(1.06944e-07)
+    assert info["cache_read_input_token_cost"] == pytest.approx(8.0208e-08)
+    assert info["output_cost_per_token"] == pytest.approx(1.0026e-06)
+    assert info["supports_function_calling"] is True
+    assert info["supports_vision"] is True
+    assert info["supports_reasoning"] is True