diff --git a/litellm/constants.py b/litellm/constants.py index 6b984c2673c..11a9574b80a 100644 --- a/litellm/constants.py +++ b/litellm/constants.py @@ -894,6 +894,7 @@ openai_compatible_endpoints: Final[list] = [ "https://api.meta.ai/v1", "https://api.cognition.ai/v1", "https://api.scx.ai/v1", + "https://api.prisminference.com/v1", "https://gigachat.devices.sberbank.ru/api/v1", ] @@ -966,6 +967,7 @@ openai_compatible_providers: Final[list] = [ "meta", # Meta Model API (Muse Spark) - JSON-configured provider "cognition", "scx-ai", + "prism", ] openai_text_completion_compatible_providers: Final[list] = [ # providers that support `/v1/completions` "together_ai", diff --git a/litellm/llms/openai_like/providers.json b/litellm/llms/openai_like/providers.json index a458a209ea9..1958f6634ff 100644 --- a/litellm/llms/openai_like/providers.json +++ b/litellm/llms/openai_like/providers.json @@ -200,5 +200,11 @@ "temperature_max": 1.99 }, "supported_endpoints": ["/v1/chat/completions"] + }, + "prism": { + "base_url": "https://api.prisminference.com/v1", + "api_key_env": "PRISM_API_KEY", + "api_base_env": "PRISM_API_BASE", + "supported_endpoints": ["/v1/chat/completions", "/v1/responses", "/v1/messages"] } } diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index 7fa09951eae..03eb89251a9 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -65063,5 +65063,29 @@ "supports_tool_choice": false, "supports_response_schema": true, "supports_vision": false + }, + "prism/deepseek-v4-flash": { + "cache_read_input_token_cost": 7e-08, + "input_cost_per_token": 1.4e-07, + "litellm_provider": "prism", + "max_input_tokens": 1000000, + "max_output_tokens": 393216, + "max_tokens": 393216, + "mode": "chat", + "output_cost_per_token": 2.8e-07, + "source": "https://prisminference.com/pricing", + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/responses", + "/v1/messages" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true } } diff --git a/litellm/provider_endpoints_support_backup.json b/litellm/provider_endpoints_support_backup.json index dbeaccdda2d..35941d357d8 100644 --- a/litellm/provider_endpoints_support_backup.json +++ b/litellm/provider_endpoints_support_backup.json @@ -1925,6 +1925,23 @@ "interactions": true } }, + "prism": { + "display_name": "Prism (`prism`)", + "url": "https://docs.litellm.ai/docs/providers/prism", + "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 + } + }, "recraft": { "display_name": "Recraft (`recraft`)", "url": "https://docs.litellm.ai/docs/providers/recraft", diff --git a/litellm/proxy/public_endpoints/provider_create_fields.json b/litellm/proxy/public_endpoints/provider_create_fields.json index cd781abee26..4ad0e527ee5 100644 --- a/litellm/proxy/public_endpoints/provider_create_fields.json +++ b/litellm/proxy/public_endpoints/provider_create_fields.json @@ -2703,6 +2703,34 @@ ], "default_model_placeholder": "gpt-3.5-turbo" }, + { + "provider": "PRISM", + "provider_display_name": "Prism", + "litellm_provider": "prism", + "credential_fields": [ + { + "key": "api_base", + "label": "API Base", + "placeholder": "https://api.prisminference.com/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": "prism/deepseek-v4-flash" + }, { "provider": "RECRAFT", "provider_display_name": "Recraft", diff --git a/litellm/types/utils.py b/litellm/types/utils.py index e3ea37dc0c8..d90233613e0 100644 --- a/litellm/types/utils.py +++ b/litellm/types/utils.py @@ -4012,6 +4012,7 @@ class LlmProviders(str, Enum): PINSTRIPES = "pinstripes" COGNITION = "cognition" SCX_AI = "scx-ai" + PRISM = "prism" DARKBLOOM = "darkbloom" META = "meta" LITELLM_AGENT = "litellm_agent" diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index 7fa09951eae..03eb89251a9 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -65063,5 +65063,29 @@ "supports_tool_choice": false, "supports_response_schema": true, "supports_vision": false + }, + "prism/deepseek-v4-flash": { + "cache_read_input_token_cost": 7e-08, + "input_cost_per_token": 1.4e-07, + "litellm_provider": "prism", + "max_input_tokens": 1000000, + "max_output_tokens": 393216, + "max_tokens": 393216, + "mode": "chat", + "output_cost_per_token": 2.8e-07, + "source": "https://prisminference.com/pricing", + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/responses", + "/v1/messages" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true } } diff --git a/provider_endpoints_support.json b/provider_endpoints_support.json index c71f4a82a4a..f1cec19edbd 100644 --- a/provider_endpoints_support.json +++ b/provider_endpoints_support.json @@ -2142,6 +2142,23 @@ "interactions": true } }, + "prism": { + "display_name": "Prism (`prism`)", + "url": "https://docs.litellm.ai/docs/providers/prism", + "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 + } + }, "recraft": { "display_name": "Recraft (`recraft`)", "url": "https://docs.litellm.ai/docs/providers/recraft", diff --git a/tests/test_litellm/llms/openai_like/test_prism_provider.py b/tests/test_litellm/llms/openai_like/test_prism_provider.py new file mode 100644 index 00000000000..ca79947aa2d --- /dev/null +++ b/tests/test_litellm/llms/openai_like/test_prism_provider.py @@ -0,0 +1,128 @@ +import json +from pathlib import Path + +import pytest + +import litellm + + +def test_prism_provider_resolution(monkeypatch: pytest.MonkeyPatch): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + monkeypatch.setenv("PRISM_API_KEY", "prism-test-key") + + model, provider, api_key, api_base = get_llm_provider( + model="prism/deepseek-v4-flash", + custom_llm_provider=None, + api_base=None, + api_key=None, + ) + + assert model == "deepseek-v4-flash" + assert provider == "prism" + assert api_key == "prism-test-key" + assert api_base == "https://api.prisminference.com/v1" + + +def test_prism_provider_keeps_explicit_credentials(monkeypatch: pytest.MonkeyPatch): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + monkeypatch.setenv("PRISM_API_KEY", "prism-env-key") + + _, provider, api_key, api_base = get_llm_provider( + model="prism/deepseek-v4-flash", + custom_llm_provider=None, + api_base="https://prism.internal.example/v1", + api_key="prism-explicit-key", + ) + + assert provider == "prism" + assert api_key == "prism-explicit-key" + assert api_base == "https://prism.internal.example/v1" + + +def test_prism_model_cost_and_capabilities(): + from litellm.cost_calculator import cost_per_token + + prompt_cost, completion_cost = cost_per_token( + model="prism/deepseek-v4-flash", + prompt_tokens=1_000_000, + completion_tokens=1_000_000, + custom_llm_provider="prism", + ) + model_info = litellm.get_model_info("prism/deepseek-v4-flash") + + assert prompt_cost == pytest.approx(0.14) + assert completion_cost == pytest.approx(0.28) + assert model_info["cache_read_input_token_cost"] == pytest.approx(7e-08) + assert model_info["max_input_tokens"] == 1_000_000 + assert model_info["max_output_tokens"] == 393_216 + assert model_info["supports_function_calling"] is True + assert model_info["supports_native_streaming"] is True + assert model_info["supports_reasoning"] is True + assert model_info["supports_response_schema"] is True + + +def test_prism_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()) + prism = next(provider for provider in providers if provider["litellm_provider"] == "prism") + + assert prism["provider"] == "PRISM" + assert prism["provider_display_name"] == "Prism" + assert prism["default_model_placeholder"] == "prism/deepseek-v4-flash" + assert {field["key"]: field["required"] for field in prism["credential_fields"]} == { + "api_base": False, + "api_key": True, + } + + +def test_prism_supported_endpoints(): + matrix_path = Path(litellm.__file__).parent / "provider_endpoints_support_backup.json" + providers = json.loads(matrix_path.read_text())["providers"] + + assert providers["prism"]["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, + } + + +def test_prism_resolves_responses_and_messages_configs(): + from litellm.llms.openai_like.json_loader import JSONProviderRegistry + from litellm.llms.openai_like.messages.transformation import ( + JSONProviderAnthropicMessagesConfig, + ) + from litellm.utils import ProviderConfigManager + + assert JSONProviderRegistry.supports_responses_api("prism") is True + responses_config = ProviderConfigManager.get_provider_responses_api_config( + provider="prism", + model="deepseek-v4-flash", + ) + messages_config = ProviderConfigManager.get_provider_anthropic_messages_config( + provider=litellm.LlmProviders.PRISM, + model="deepseek-v4-flash", + ) + + assert responses_config is not None + assert responses_config.custom_llm_provider == "prism" + assert isinstance(messages_config, JSONProviderAnthropicMessagesConfig) + assert ( + messages_config.get_complete_url( + api_base=None, + api_key="prism-test-key", + model="deepseek-v4-flash", + optional_params={}, + litellm_params={}, + ) + == "https://api.prisminference.com/v1/messages" + )