diff --git a/README.md b/README.md index 4feb6e00ffa..36e763802c9 100644 --- a/README.md +++ b/README.md @@ -411,6 +411,7 @@ Set `LITELLM_PROXY_API_BASE` and `LITELLM_PROXY_API_KEY` and every model call th | [Sambanova (`sambanova`)](https://docs.litellm.ai/docs/providers/sambanova) | ✅ | ✅ | ✅ | | | | | | | | | [ScaleDown (`scaledown`)](https://docs.litellm.ai/docs/providers/scaledown) | ✅ | | | | | | | | | | | [Snowflake (`snowflake`)](https://docs.litellm.ai/docs/providers/snowflake) | ✅ | ✅ | ✅ | | | | | | | | +| [STACKIT (`stackit`)](https://docs.litellm.ai/docs/providers/stackit) | ✅ | ✅ | ✅ | | | | | | | | | [Strands Decider (`strands_decider`)](https://docs.litellm.ai/docs/providers) | | | | | | | | | | | | [Text Completion Codestral (`text-completion-codestral`)](https://docs.litellm.ai/docs/providers/codestral) | ✅ | ✅ | ✅ | | | | | | | | | [Text Completion OpenAI (`text-completion-openai`)](https://docs.litellm.ai/docs/providers/text_completion_openai) | ✅ | ✅ | ✅ | | | ✅ | ✅ | ✅ | ✅ | | diff --git a/litellm/constants.py b/litellm/constants.py index d1daf2ee2d1..46411b6ad72 100644 --- a/litellm/constants.py +++ b/litellm/constants.py @@ -1060,6 +1060,7 @@ openai_compatible_endpoints: Final[list[str]] = [ "https://api.scx.ai/v1", "https://api.prisminference.com/v1", "https://api.reka.ai/v1", + "https://api.openai-compat.model-serving.eu01.onstackit.cloud/v1", "https://gigachat.devices.sberbank.ru/api/v1", "https://api.openinfer.ai/v1", ] @@ -1141,6 +1142,7 @@ openai_compatible_providers: Final[list[str]] = [ "prism", "sail", "reka", + "stackit", ] OPENAI_AUDIO_TRANSCRIPTION_PROVIDERS: Final = frozenset({"openai"} | frozenset(openai_compatible_providers)) diff --git a/litellm/llms/openai_like/providers.json b/litellm/llms/openai_like/providers.json index 36337fddbdc..535feb22731 100644 --- a/litellm/llms/openai_like/providers.json +++ b/litellm/llms/openai_like/providers.json @@ -249,5 +249,11 @@ "/v1/responses", "/v1/messages" ] + }, + "stackit": { + "base_url": "https://api.openai-compat.model-serving.eu01.onstackit.cloud/v1", + "api_key_env": "STACKIT_API_KEY", + "api_base_env": "STACKIT_API_BASE", + "supported_endpoints": ["/v1/chat/completions"] } } diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index 9a7d88b8eda..85c3572be2a 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -67899,6 +67899,94 @@ "supports_reasoning": true, "supports_vision": false }, + "stackit/Qwen/Qwen3-VL-235B-A22B-Instruct-FP8": { + "input_cost_per_token": 1.5e-06, + "litellm_provider": "stackit", + "max_input_tokens": 200000, + "max_output_tokens": 16384, + "max_tokens": 16384, + "mode": "chat", + "output_cost_per_token": 1.75e-06, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true, + "supports_vision": true + }, + "stackit/Qwen/Qwen3.8-27B": { + "input_cost_per_token": 4.5e-07, + "litellm_provider": "stackit", + "max_input_tokens": 262144, + "max_output_tokens": 16384, + "max_tokens": 16384, + "mode": "chat", + "output_cost_per_token": 6.5e-07, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_vision": true + }, + "stackit/Qwen/Qwen3.6-27B": { + "deprecation_date": "2026-12-08", + "input_cost_per_token": 4.5e-07, + "litellm_provider": "stackit", + "max_input_tokens": 262144, + "max_output_tokens": 16384, + "max_tokens": 16384, + "mode": "chat", + "output_cost_per_token": 6.5e-07, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_vision": true + }, + "stackit/cortecs/Llama-3.3-70B-Instruct-FP8-Dynamic": { + "input_cost_per_token": 4.5e-07, + "litellm_provider": "stackit", + "max_input_tokens": 131072, + "max_output_tokens": 4096, + "max_tokens": 4096, + "mode": "chat", + "output_cost_per_token": 6.5e-07, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true + }, + "stackit/openai/gpt-oss-120b": { + "input_cost_per_token": 4.5e-07, + "litellm_provider": "stackit", + "max_input_tokens": 131072, + "max_output_tokens": 8192, + "max_tokens": 8192, + "mode": "chat", + "output_cost_per_token": 6.5e-07, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_response_schema": true + }, + "stackit/google/gemma-4-31B-it": { + "input_cost_per_token": 4.5e-07, + "litellm_provider": "stackit", + "max_input_tokens": 262144, + "max_output_tokens": 4096, + "max_tokens": 4096, + "mode": "chat", + "output_cost_per_token": 6.5e-07, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_vision": true + }, + "stackit/openai/gpt-oss-20b": { + "input_cost_per_token": 1.5e-07, + "litellm_provider": "stackit", + "max_input_tokens": 131072, + "max_output_tokens": 8192, + "max_tokens": 8192, + "mode": "chat", + "output_cost_per_token": 2.5e-07, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true, + "supports_reasoning": true + }, "azure_ai/kimi-k2.7-code": { "cache_read_input_token_cost": 1.9e-07, "deprecation_date": "2027-04-16", diff --git a/litellm/provider_endpoints_support_backup.json b/litellm/provider_endpoints_support_backup.json index f95b553c2b4..2bc29f2d464 100644 --- a/litellm/provider_endpoints_support_backup.json +++ b/litellm/provider_endpoints_support_backup.json @@ -2225,6 +2225,24 @@ "interactions": true } }, + "stackit": { + "display_name": "STACKIT (`stackit`)", + "url": "https://docs.litellm.ai/docs/providers/stackit", + "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 + } + }, "synthetic": { "display_name": "Synthetic (`synthetic`)", "endpoints": { diff --git a/litellm/proxy/public_endpoints/provider_create_fields.json b/litellm/proxy/public_endpoints/provider_create_fields.json index 19ee465c7dc..0a011bed421 100644 --- a/litellm/proxy/public_endpoints/provider_create_fields.json +++ b/litellm/proxy/public_endpoints/provider_create_fields.json @@ -3481,6 +3481,34 @@ ], "default_model_placeholder": "soniox/stt-async-v5" }, + { + "provider": "STACKIT", + "provider_display_name": "STACKIT", + "litellm_provider": "stackit", + "credential_fields": [ + { + "key": "api_base", + "label": "API Base", + "placeholder": "https://api.openai-compat.model-serving.eu01.onstackit.cloud/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": "stackit/openai/gpt-oss-120b" + }, { "provider": "StrandsDecider", "provider_display_name": "Strands Decider", diff --git a/litellm/types/utils.py b/litellm/types/utils.py index cc7ef851111..c5d71503fbd 100644 --- a/litellm/types/utils.py +++ b/litellm/types/utils.py @@ -4586,6 +4586,7 @@ class LlmProviders(str, Enum): META = "meta" SAIL = "sail" REKA = "reka" + STACKIT = "stackit" LITELLM_AGENT = "litellm_agent" CURSOR = "cursor" BEDROCK_MANTLE = "bedrock_mantle" diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index 9a7d88b8eda..85c3572be2a 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -67899,6 +67899,94 @@ "supports_reasoning": true, "supports_vision": false }, + "stackit/Qwen/Qwen3-VL-235B-A22B-Instruct-FP8": { + "input_cost_per_token": 1.5e-06, + "litellm_provider": "stackit", + "max_input_tokens": 200000, + "max_output_tokens": 16384, + "max_tokens": 16384, + "mode": "chat", + "output_cost_per_token": 1.75e-06, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true, + "supports_vision": true + }, + "stackit/Qwen/Qwen3.8-27B": { + "input_cost_per_token": 4.5e-07, + "litellm_provider": "stackit", + "max_input_tokens": 262144, + "max_output_tokens": 16384, + "max_tokens": 16384, + "mode": "chat", + "output_cost_per_token": 6.5e-07, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_vision": true + }, + "stackit/Qwen/Qwen3.6-27B": { + "deprecation_date": "2026-12-08", + "input_cost_per_token": 4.5e-07, + "litellm_provider": "stackit", + "max_input_tokens": 262144, + "max_output_tokens": 16384, + "max_tokens": 16384, + "mode": "chat", + "output_cost_per_token": 6.5e-07, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_vision": true + }, + "stackit/cortecs/Llama-3.3-70B-Instruct-FP8-Dynamic": { + "input_cost_per_token": 4.5e-07, + "litellm_provider": "stackit", + "max_input_tokens": 131072, + "max_output_tokens": 4096, + "max_tokens": 4096, + "mode": "chat", + "output_cost_per_token": 6.5e-07, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true + }, + "stackit/openai/gpt-oss-120b": { + "input_cost_per_token": 4.5e-07, + "litellm_provider": "stackit", + "max_input_tokens": 131072, + "max_output_tokens": 8192, + "max_tokens": 8192, + "mode": "chat", + "output_cost_per_token": 6.5e-07, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_response_schema": true + }, + "stackit/google/gemma-4-31B-it": { + "input_cost_per_token": 4.5e-07, + "litellm_provider": "stackit", + "max_input_tokens": 262144, + "max_output_tokens": 4096, + "max_tokens": 4096, + "mode": "chat", + "output_cost_per_token": 6.5e-07, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_vision": true + }, + "stackit/openai/gpt-oss-20b": { + "input_cost_per_token": 1.5e-07, + "litellm_provider": "stackit", + "max_input_tokens": 131072, + "max_output_tokens": 8192, + "max_tokens": 8192, + "mode": "chat", + "output_cost_per_token": 2.5e-07, + "source": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/", + "supports_function_calling": true, + "supports_reasoning": true + }, "azure_ai/kimi-k2.7-code": { "cache_read_input_token_cost": 1.9e-07, "deprecation_date": "2027-04-16", diff --git a/provider_endpoints_support.json b/provider_endpoints_support.json index d65da3f41fd..0de254bd157 100644 --- a/provider_endpoints_support.json +++ b/provider_endpoints_support.json @@ -2513,6 +2513,24 @@ "rerank": false } }, + "stackit": { + "display_name": "STACKIT (`stackit`)", + "url": "https://docs.litellm.ai/docs/providers/stackit", + "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 + } + }, "synthetic": { "display_name": "Synthetic (`synthetic`)", "endpoints": { diff --git a/tests/unit/llms/openai_like/test_stackit_provider.py b/tests/unit/llms/openai_like/test_stackit_provider.py new file mode 100644 index 00000000000..73b36fbc584 --- /dev/null +++ b/tests/unit/llms/openai_like/test_stackit_provider.py @@ -0,0 +1,243 @@ +import json +from pathlib import Path +from typing import Final + +import pytest +import respx + +import litellm +from litellm.caching.llm_caching_handler import LLMClientCache + +_BASE_URL: Final = "https://api.openai-compat.model-serving.eu01.onstackit.cloud/v1" +_REPO_ROOT: Final = Path(litellm.__file__).parent.parent + +_CHAT_COMPLETION: Final = { + "id": "chatcmpl_stackit", + "object": "chat.completion", + "created": 1_790_000_000, + "model": "openai/gpt-oss-120b", + "choices": [ + { + "index": 0, + "message": {"role": "assistant", "content": "Hello from STACKIT"}, + "finish_reason": "stop", + } + ], + "usage": {"prompt_tokens": 4, "completion_tokens": 3, "total_tokens": 7}, +} + + +def test_stackit_is_a_registered_provider(): + assert litellm.LlmProviders.STACKIT.value == "stackit" + assert "stackit" in litellm.provider_list + assert "stackit" in litellm.constants.openai_compatible_providers + assert _BASE_URL in litellm.constants.openai_compatible_endpoints + + +def test_stackit_provider_resolution(monkeypatch: pytest.MonkeyPatch): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + monkeypatch.setenv("STACKIT_API_KEY", "stackit-test-key") + monkeypatch.delenv("STACKIT_API_BASE", raising=False) + + model, provider, api_key, api_base = get_llm_provider( + model="stackit/openai/gpt-oss-120b", + custom_llm_provider=None, + api_base=None, + api_key=None, + ) + + assert model == "openai/gpt-oss-120b" + assert provider == "stackit" + assert api_key == "stackit-test-key" + assert api_base == _BASE_URL + + +def test_stackit_provider_keeps_explicit_credentials(monkeypatch: pytest.MonkeyPatch): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + monkeypatch.setenv("STACKIT_API_KEY", "stackit-env-key") + monkeypatch.setenv("STACKIT_API_BASE", "https://stackit.env.example/v1") + + _, provider, api_key, api_base = get_llm_provider( + model="stackit/openai/gpt-oss-120b", + custom_llm_provider=None, + api_base="https://stackit.internal.example/v1", + api_key="stackit-explicit-key", + ) + + assert provider == "stackit" + assert api_key == "stackit-explicit-key" + assert api_base == "https://stackit.internal.example/v1" + + +def test_stackit_api_base_env_overrides_default(monkeypatch: pytest.MonkeyPatch): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + monkeypatch.setenv("STACKIT_API_KEY", "stackit-env-key") + monkeypatch.setenv("STACKIT_API_BASE", "https://stackit.env.example/v1") + + _, provider, _, api_base = get_llm_provider( + model="stackit/openai/gpt-oss-120b", + custom_llm_provider=None, + api_base=None, + api_key=None, + ) + + assert provider == "stackit" + assert api_base == "https://stackit.env.example/v1" + + +def test_stackit_api_base_autodetects_provider(monkeypatch: pytest.MonkeyPatch): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + monkeypatch.setenv("STACKIT_API_KEY", "stackit-env-key") + + # STACKIT model ids carry their own org prefix (openai/, google/, Qwen/, ...). "openai/..." on its own + # would be routed to OpenAI, so autodetection is only meaningful for ids without a provider-like prefix. + model, provider, api_key, api_base = get_llm_provider( + model="Qwen/Qwen3.8-27B", + custom_llm_provider=None, + api_base=_BASE_URL, + api_key=None, + ) + + assert model == "Qwen/Qwen3.8-27B" + assert provider == "stackit" + assert api_key == "stackit-env-key" + assert api_base == _BASE_URL + + +def test_stackit_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()) + stackit = next(provider for provider in providers if provider["litellm_provider"] == "stackit") + + assert stackit["provider"] == "STACKIT" + assert stackit["provider_display_name"] == "STACKIT" + assert stackit["default_model_placeholder"] == "stackit/openai/gpt-oss-120b" + assert {field["key"]: field["required"] for field in stackit["credential_fields"]} == { + "api_base": False, + "api_key": True, + } + + +def test_stackit_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 = _REPO_ROOT / "provider_endpoints_support.json" + + assert json.loads(backup_path.read_text())["providers"]["stackit"]["endpoints"] == expected + assert json.loads(root_path.read_text())["providers"]["stackit"]["endpoints"] == expected + + +def test_stackit_cost_map_entries(): + root = json.loads((_REPO_ROOT / "model_prices_and_context_window.json").read_text()) + backup = json.loads((Path(litellm.__file__).parent / "model_prices_and_context_window_backup.json").read_text()) + + for cost_map in (root, backup): + stackit = {k: v for k, v in cost_map.items() if v.get("litellm_provider") == "stackit"} + assert stackit, "no stackit entries in cost map" + assert all(key.startswith("stackit/") for key in stackit) + assert {v["mode"] for v in stackit.values()} == {"chat"} + + # STACKIT price tiers (EUR per 1M tokens): standard 0.15/0.25, plus 0.45/0.65, premium 1.50/1.75 + gpt_oss_20b = stackit["stackit/openai/gpt-oss-20b"] + assert gpt_oss_20b["input_cost_per_token"] == pytest.approx(0.15e-6) + assert gpt_oss_20b["output_cost_per_token"] == pytest.approx(0.25e-6) + + gpt_oss_120b = stackit["stackit/openai/gpt-oss-120b"] + assert gpt_oss_120b["input_cost_per_token"] == pytest.approx(0.45e-6) + assert gpt_oss_120b["output_cost_per_token"] == pytest.approx(0.65e-6) + assert gpt_oss_120b["supports_reasoning"] is True + assert gpt_oss_120b["supports_function_calling"] is True + + qwen_vl = stackit["stackit/Qwen/Qwen3-VL-235B-A22B-Instruct-FP8"] + assert qwen_vl["input_cost_per_token"] == pytest.approx(1.50e-6) + assert qwen_vl["output_cost_per_token"] == pytest.approx(1.75e-6) + assert qwen_vl["supports_vision"] is True + + +def test_stackit_chat_completion_request(): + with respx.mock() as upstream: + route: Final = upstream.post(f"{_BASE_URL}/chat/completions").respond(200, json=_CHAT_COMPLETION) + response: Final = litellm.completion( + model="stackit/openai/gpt-oss-120b", + messages=[{"role": "user", "content": "Say hello"}], + api_key="stackit-test-key", + ) + + request: Final = route.calls.last.request + body: Final = json.loads(request.content) + assert route.call_count == 1 + assert str(request.url) == f"{_BASE_URL}/chat/completions" + assert request.headers["authorization"] == "Bearer stackit-test-key" + assert body["model"] == "openai/gpt-oss-120b" + assert body["messages"] == [{"role": "user", "content": "Say hello"}] + assert response.choices[0].message.content == "Hello from STACKIT" + + +def test_stackit_chat_completion_cost_uses_cost_map(): + with respx.mock() as upstream: + upstream.post(f"{_BASE_URL}/chat/completions").respond(200, json=_CHAT_COMPLETION) + response: Final = litellm.completion( + model="stackit/openai/gpt-oss-120b", + messages=[{"role": "user", "content": "Say hello"}], + api_key="stackit-test-key", + ) + + # 4 prompt tokens * 0.45e-6 + 3 completion tokens * 0.65e-6 + assert response._hidden_params["response_cost"] == pytest.approx(4 * 0.45e-6 + 3 * 0.65e-6) + + +def test_stackit_responses_request_is_bridged_to_chat_completions(): + with respx.mock() as upstream: + route: Final = upstream.post(f"{_BASE_URL}/chat/completions").respond(200, json=_CHAT_COMPLETION) + response: Final = litellm.responses( + model="stackit/openai/gpt-oss-120b", + input="Say hello", + api_key="stackit-test-key", + ) + + request: Final = route.calls.last.request + body: Final = json.loads(request.content) + assert route.call_count == 1 + assert request.headers["authorization"] == "Bearer stackit-test-key" + assert body["model"] == "openai/gpt-oss-120b" + assert body["messages"] == [{"role": "user", "content": "Say hello"}] + assert response.output[0].content[0].text == "Hello from STACKIT" + + +@pytest.mark.asyncio +async def test_stackit_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(f"{_BASE_URL}/chat/completions").respond(200, json=_CHAT_COMPLETION) + response: Final = await litellm.anthropic.messages.acreate( + model="stackit/openai/gpt-oss-120b", + messages=[{"role": "user", "content": "Say hello"}], + max_tokens=32, + api_key="stackit-test-key", + ) + + request: Final = route.calls.last.request + body: Final = json.loads(request.content) + assert route.call_count == 1 + assert request.headers["authorization"] == "Bearer stackit-test-key" + assert body["model"] == "openai/gpt-oss-120b" + assert body["max_tokens"] == 32 + assert response["content"][0]["text"] == "Hello from STACKIT"