From d206568d529855b22fb49ec5557fbb7ee0bfe990 Mon Sep 17 00:00:00 2001 From: Codonyat Date: Sun, 13 Sep 2026 12:39:05 +0200 Subject: [PATCH] feat(providers): add Gondola as a JSON-configured OpenAI-compatible provider Gondola (https://gondola-ai.com) is an OpenAI-compatible gateway in front of Venice AI's model catalog, paid in USDC. This registers it the same way the other JSON-configured providers are registered: - providers.json entry (base url, GONDOLA_API_KEY, GONDOLA_API_BASE, max_completion_tokens -> max_tokens), plus the LlmProviders enum member, openai_compatible_providers and openai_compatible_endpoints entries so a caller-supplied api_base resolves to the provider. - supported_endpoints opts into the native Anthropic Messages passthrough: the gateway serves /v1/messages natively and accepts a Bearer key, so LiteLLM forwards Anthropic payloads untranslated like it does for meta. Embeddings are deliberately NOT declared: the gateway serves them, but litellm.embedding() has no dispatch for JSON-configured providers. - provider_endpoints_support.json and its backup matrix, and the Add Model dropdown entry in provider_create_fields.json (required by test_every_backend_provider_is_listed_in_add_model_or_frozen_as_unlisted). - Dashboard logo and provider maps, with unit tests. - No model_prices_and_context_window.json rows: Gondola resells ~120 text models at market prices, like nano-gpt, poe, chutes and veniceai, which also carry none. Rebuilt on the current staging branch; the July branch predates the dropdown parity test and the generic api_base resolution for JSON providers. Co-Authored-By: Claude Fable 5.1 Claude-Session: https://claude.ai/code/session_019wZU1Uoj94ZECV4VC1hesL --- litellm/constants.py | 2 + litellm/llms/openai_like/providers.json | 9 + .../provider_endpoints_support_backup.json | 17 ++ .../provider_create_fields.json | 28 ++ litellm/types/utils.py | 1 + provider_endpoints_support.json | 17 ++ .../test_get_llm_provider_endpoint_match.py | 33 +++ .../llms/openai_like/test_gondola_provider.py | 243 ++++++++++++++++++ .../public/assets/logos/gondola.svg | 11 + .../components/provider_info_helpers.test.tsx | 15 ++ .../src/components/provider_info_helpers.tsx | 5 + 11 files changed, 381 insertions(+) create mode 100644 tests/test_litellm/llms/openai_like/test_gondola_provider.py create mode 100644 ui/litellm-dashboard/public/assets/logos/gondola.svg diff --git a/litellm/constants.py b/litellm/constants.py index 5751e6e46af..e80151249a5 100644 --- a/litellm/constants.py +++ b/litellm/constants.py @@ -883,6 +883,7 @@ openai_compatible_endpoints: Final[list] = [ "https://nano-gpt.com/api/v1", "https://api.poe.com/v1", "https://llm.chutes.ai/v1/", + "https://api.gondola-ai.com/v1", "https://api.v0.dev/v1", "https://api.morphllm.com/v1", "https://api.lambda.ai/v1", @@ -941,6 +942,7 @@ openai_compatible_providers: Final[list] = [ "nano-gpt", # Nano-GPT - JSON-configured provider "poe", # Poe - JSON-configured provider "chutes", # Chutes - JSON-configured provider + "gondola", # Gondola - JSON-configured provider "parasail", # Parasail - JSON-configured provider "libertai", # LibertAI - JSON-configured provider "featherless_ai", diff --git a/litellm/llms/openai_like/providers.json b/litellm/llms/openai_like/providers.json index fe10293c420..6bf40c7767b 100644 --- a/litellm/llms/openai_like/providers.json +++ b/litellm/llms/openai_like/providers.json @@ -65,6 +65,15 @@ "max_completion_tokens": "max_tokens" } }, + "gondola": { + "base_url": "https://api.gondola-ai.com/v1", + "api_key_env": "GONDOLA_API_KEY", + "api_base_env": "GONDOLA_API_BASE", + "param_mappings": { + "max_completion_tokens": "max_tokens" + }, + "supported_endpoints": ["/v1/chat/completions", "/v1/messages"] + }, "abliteration": { "base_url": "https://api.abliteration.ai/v1", "api_key_env": "ABLITERATION_API_KEY" diff --git a/litellm/provider_endpoints_support_backup.json b/litellm/provider_endpoints_support_backup.json index dbeaccdda2d..f3d025242df 100644 --- a/litellm/provider_endpoints_support_backup.json +++ b/litellm/provider_endpoints_support_backup.json @@ -1098,6 +1098,23 @@ "interactions": true } }, + "gondola": { + "display_name": "Gondola (`gondola`)", + "url": "https://docs.litellm.ai/docs/providers/gondola", + "endpoints": { + "chat_completions": true, + "messages": true, + "responses": false, + "embeddings": false, + "image_generations": false, + "audio_transcriptions": false, + "audio_speech": false, + "moderations": false, + "batches": false, + "rerank": false, + "a2a": false + } + }, "vertex_ai": { "display_name": "Google - Vertex AI (`vertex_ai`)", "url": "https://docs.litellm.ai/docs/providers/vertex", diff --git a/litellm/proxy/public_endpoints/provider_create_fields.json b/litellm/proxy/public_endpoints/provider_create_fields.json index cd781abee26..99f67eddca6 100644 --- a/litellm/proxy/public_endpoints/provider_create_fields.json +++ b/litellm/proxy/public_endpoints/provider_create_fields.json @@ -1499,6 +1499,34 @@ ], "default_model_placeholder": "gpt-3.5-turbo" }, + { + "provider": "GONDOLA", + "provider_display_name": "Gondola", + "litellm_provider": "gondola", + "credential_fields": [ + { + "key": "api_base", + "label": "API Base", + "placeholder": "https://api.gondola-ai.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": "gondola/deepseek-v3.2" + }, { "provider": "Google_AI_Studio", "provider_display_name": "Google AI Studio", diff --git a/litellm/types/utils.py b/litellm/types/utils.py index 00c55b35182..f797dadb8b2 100644 --- a/litellm/types/utils.py +++ b/litellm/types/utils.py @@ -4038,6 +4038,7 @@ class LlmProviders(str, Enum): NANOGPT = "nano-gpt" POE = "poe" CHUTES = "chutes" + GONDOLA = "gondola" NEOSANTARA = "neosantara" PARASAIL = "parasail" XIAOMI_MIMO = "xiaomi_mimo" diff --git a/provider_endpoints_support.json b/provider_endpoints_support.json index c71f4a82a4a..68976b5c2fc 100644 --- a/provider_endpoints_support.json +++ b/provider_endpoints_support.json @@ -1195,6 +1195,23 @@ "interactions": true } }, + "gondola": { + "display_name": "Gondola (`gondola`)", + "url": "https://docs.litellm.ai/docs/providers/gondola", + "endpoints": { + "chat_completions": true, + "messages": true, + "responses": false, + "embeddings": false, + "image_generations": false, + "audio_transcriptions": false, + "audio_speech": false, + "moderations": false, + "batches": false, + "rerank": false, + "a2a": false + } + }, "vertex_ai": { "display_name": "Google - Vertex AI (`vertex_ai`)", "url": "https://docs.litellm.ai/docs/providers/vertex", diff --git a/tests/test_litellm/litellm_core_utils/test_get_llm_provider_endpoint_match.py b/tests/test_litellm/litellm_core_utils/test_get_llm_provider_endpoint_match.py index 419ca104bb1..912389b0fff 100644 --- a/tests/test_litellm/litellm_core_utils/test_get_llm_provider_endpoint_match.py +++ b/tests/test_litellm/litellm_core_utils/test_get_llm_provider_endpoint_match.py @@ -207,3 +207,36 @@ class TestGigachatApiBaseResolvesProvider: assert dynamic_api_key == "gigachat-key-from-env" assert returned_api_base == "https://gigachat.devices.sberbank.ru/api/v1" assert model == "GigaChat-2" + + +class TestGondolaApiBaseResolution: + """ + ``api_base`` pointing at Gondola's gateway resolves to the ``gondola`` + provider through the generic JSON-provider branch, and a caller-supplied + key wins over the stored env secret so another provider's credential is + never sent to Gondola. + """ + + def test_gondola_api_base_resolves_with_env_key(self, monkeypatch): + monkeypatch.setenv("GONDOLA_API_KEY", "gnd_from_env") + + _, provider, dynamic_api_key, api_base = get_llm_provider( + model="some-model", + api_base="https://api.gondola-ai.com/v1", + ) + + assert provider == "gondola" + assert dynamic_api_key == "gnd_from_env" + assert api_base == "https://api.gondola-ai.com/v1" + + def test_caller_supplied_key_is_not_replaced_by_env_secret(self, monkeypatch): + monkeypatch.setenv("GONDOLA_API_KEY", "gnd_from_env") + + _, provider, dynamic_api_key, _ = get_llm_provider( + model="some-model", + api_base="https://api.gondola-ai.com/v1", + api_key="gnd_caller-supplied", + ) + + assert provider == "gondola" + assert dynamic_api_key == "gnd_caller-supplied" diff --git a/tests/test_litellm/llms/openai_like/test_gondola_provider.py b/tests/test_litellm/llms/openai_like/test_gondola_provider.py new file mode 100644 index 00000000000..19d05cf51d2 --- /dev/null +++ b/tests/test_litellm/llms/openai_like/test_gondola_provider.py @@ -0,0 +1,243 @@ +""" +Tests for the Gondola provider configuration and integration. + +Gondola (https://gondola-ai.com) is an OpenAI-compatible gateway in front of +Venice AI's model catalog. Besides ``/v1/chat/completions`` it natively serves +the Anthropic Messages API at ``/v1/messages``, so the JSON provider opts into +the untranslated passthrough. +""" + +import json +from pathlib import Path + +import litellm + +REPO_ROOT = Path(__file__).parents[4] + + +class TestGondolaProviderConfig: + def test_gondola_in_provider_list(self): + from litellm import LlmProviders + + assert hasattr(LlmProviders, "GONDOLA") + assert LlmProviders.GONDOLA.value == "gondola" + assert "gondola" in litellm.provider_list + + def test_gondola_json_config_exists(self): + from litellm.llms.openai_like.json_loader import JSONProviderRegistry + + assert JSONProviderRegistry.exists("gondola") + + gondola = JSONProviderRegistry.get("gondola") + assert gondola is not None + assert gondola.base_url == "https://api.gondola-ai.com/v1" + assert gondola.api_key_env == "GONDOLA_API_KEY" + assert gondola.api_base_env == "GONDOLA_API_BASE" + assert gondola.param_mappings.get("max_completion_tokens") == "max_tokens" + assert set(gondola.supported_endpoints) == {"/v1/chat/completions", "/v1/messages"} + + def test_gondola_in_openai_compatible_lists(self): + from litellm.constants import openai_compatible_endpoints, openai_compatible_providers + + assert "gondola" in openai_compatible_providers + assert "https://api.gondola-ai.com/v1" in openai_compatible_endpoints + + def test_gondola_provider_resolution(self): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + model, provider, api_key, api_base = get_llm_provider( + model="gondola/deepseek-v3.2", + custom_llm_provider=None, + api_base=None, + api_key=None, + ) + + assert model == "deepseek-v3.2" + assert provider == "gondola" + assert api_base == "https://api.gondola-ai.com/v1" + + def test_gondola_api_base_override(self): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + model, provider, api_key, api_base = get_llm_provider( + model="gondola/deepseek-v3.2", + custom_llm_provider=None, + api_base="https://proxy.example.com/v1", + api_key="gnd_test", + ) + + assert provider == "gondola" + assert api_base == "https://proxy.example.com/v1" + assert api_key == "gnd_test" + + def test_gondola_api_base_from_env(self, monkeypatch): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + monkeypatch.setenv("GONDOLA_API_BASE", "https://proxy.example.com/v1") + + _, provider, _, api_base = get_llm_provider( + model="gondola/deepseek-v3.2", + custom_llm_provider=None, + api_base=None, + api_key=None, + ) + + assert provider == "gondola" + assert api_base == "https://proxy.example.com/v1" + + def test_gondola_url_autodetection(self): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + model, provider, api_key, api_base = get_llm_provider( + model="deepseek-v3.2", + custom_llm_provider=None, + api_base="https://api.gondola-ai.com/v1", + api_key=None, + ) + + assert provider == "gondola" + assert api_base == "https://api.gondola-ai.com/v1" + + def test_gondola_max_completion_tokens_mapped(self): + from litellm.llms.openai_like.dynamic_config import create_config_class + from litellm.llms.openai_like.json_loader import JSONProviderRegistry + + provider = JSONProviderRegistry.get("gondola") + assert provider is not None + config = create_config_class(provider)() + + optional_params = config.map_openai_params( + non_default_params={"max_completion_tokens": 256}, + optional_params={}, + model="deepseek-v3.2", + drop_params=False, + ) + assert optional_params["max_tokens"] == 256 + assert "max_completion_tokens" not in optional_params + + def test_gondola_router_config(self): + from litellm import Router + + router = Router( + model_list=[ + { + "model_name": "gondola-chat", + "litellm_params": { + "model": "gondola/deepseek-v3.2", + "api_key": "gnd_test", + }, + } + ] + ) + + assert len(router.model_list) == 1 + assert router.model_list[0]["model_name"] == "gondola-chat" + + +class TestGondolaAnthropicMessages: + def test_gondola_resolves_native_messages_config(self): + from litellm.llms.openai_like.messages.transformation import ( + JSONProviderAnthropicMessagesConfig, + ) + + cfg = litellm.ProviderConfigManager.get_provider_anthropic_messages_config( + model="deepseek-v3.2", provider=litellm.LlmProviders.GONDOLA + ) + assert isinstance(cfg, JSONProviderAnthropicMessagesConfig) + + def test_complete_url_targets_gondola_messages(self): + from litellm.llms.openai_like.json_loader import JSONProviderRegistry + from litellm.llms.openai_like.messages.transformation import ( + JSONProviderAnthropicMessagesConfig, + ) + + provider = JSONProviderRegistry.get("gondola") + assert provider is not None + cfg = JSONProviderAnthropicMessagesConfig(provider) + + url = cfg.get_complete_url( + api_base=None, + api_key="gnd_test", + model="deepseek-v3.2", + optional_params={}, + litellm_params={}, + ) + assert url == "https://api.gondola-ai.com/v1/messages" + + override_url = cfg.get_complete_url( + api_base="https://proxy.example.com/v1", + api_key="gnd_test", + model="deepseek-v3.2", + optional_params={}, + litellm_params={}, + ) + assert override_url == "https://proxy.example.com/v1/messages" + + def test_api_key_resolved_from_env(self, monkeypatch): + from litellm.llms.openai_like.json_loader import JSONProviderRegistry + from litellm.llms.openai_like.messages.transformation import ( + JSONProviderAnthropicMessagesConfig, + ) + + monkeypatch.setenv("GONDOLA_API_KEY", "gnd_env_key") + provider = JSONProviderRegistry.get("gondola") + assert provider is not None + cfg = JSONProviderAnthropicMessagesConfig(provider) + + headers, _ = cfg.validate_anthropic_messages_environment( + headers={}, + model="deepseek-v3.2", + messages=[{"role": "user", "content": "hi"}], + optional_params={}, + litellm_params={}, + api_key=None, + api_base=None, + ) + assert headers["authorization"] == "Bearer gnd_env_key" + assert headers["anthropic-version"] == "2023-06-01" + + +class TestGondolaEndpointMatrix: + @staticmethod + def _providers(path_parts): + with open(REPO_ROOT.joinpath(*path_parts)) as f: + return json.load(f)["providers"] + + def test_gondola_listed_in_supported_endpoints_and_backup(self): + root = self._providers(("provider_endpoints_support.json",)) + backup = self._providers(("litellm", "provider_endpoints_support_backup.json")) + + assert "gondola" in root, "gondola missing from provider_endpoints_support.json" + assert "gondola" in backup, "gondola missing from the backup matrix" + assert backup["gondola"] == root["gondola"] + + endpoints = root["gondola"]["endpoints"] + assert endpoints["chat_completions"] is True + assert endpoints["messages"] is True + # LiteLLM has no embedding dispatch for JSON-configured providers, so the + # matrix must not advertise one even though the gateway serves it. + assert endpoints["embeddings"] is False + assert endpoints["responses"] is False + + +class TestGondolaDashboardRegistration: + @staticmethod + def _provider_create_fields(): + path = Path(litellm.__file__).parent / "proxy" / "public_endpoints" / "provider_create_fields.json" + with open(path) as f: + return json.load(f) + + def test_gondola_is_selectable_in_the_add_model_form(self): + entries = [e for e in self._provider_create_fields() if e["litellm_provider"] == "gondola"] + assert len(entries) == 1, "gondola must appear exactly once in provider_create_fields.json" + + entry = entries[0] + assert entry["provider"] == "GONDOLA" + assert entry["provider_display_name"] == "Gondola" + assert entry["default_model_placeholder"].startswith("gondola/") + + fields = {f["key"]: f for f in entry["credential_fields"]} + assert fields["api_key"]["required"] is True + assert fields["api_key"]["field_type"] == "password" + assert fields["api_base"]["required"] is False + assert fields["api_base"]["placeholder"] == "https://api.gondola-ai.com/v1" diff --git a/ui/litellm-dashboard/public/assets/logos/gondola.svg b/ui/litellm-dashboard/public/assets/logos/gondola.svg new file mode 100644 index 00000000000..4598dfbf9b3 --- /dev/null +++ b/ui/litellm-dashboard/public/assets/logos/gondola.svg @@ -0,0 +1,11 @@ + + + + + + + + + + + diff --git a/ui/litellm-dashboard/src/components/provider_info_helpers.test.tsx b/ui/litellm-dashboard/src/components/provider_info_helpers.test.tsx index 4c68e302267..d58bb756e8b 100644 --- a/ui/litellm-dashboard/src/components/provider_info_helpers.test.tsx +++ b/ui/litellm-dashboard/src/components/provider_info_helpers.test.tsx @@ -62,6 +62,17 @@ describe("provider_info_helpers", () => { expect(result.logo).toBe(providerLogoMap[Providers.Groq]); }); + it("should map gondola slug and GONDOLA enum key to the Gondola display name and logo", () => { + const fromSlug = getProviderLogoAndName("gondola"); + expect(fromSlug.displayName).toBe(Providers.GONDOLA); + expect(fromSlug.logo).toBe(providerLogoMap[Providers.GONDOLA]); + expect(fromSlug.logo).toBeTruthy(); + + const fromEnumKey = getProviderLogoAndName("GONDOLA"); + expect(fromEnumKey.displayName).toBe(Providers.GONDOLA); + expect(fromEnumKey.logo).toBe(providerLogoMap[Providers.GONDOLA]); + }); + it("should map scx-ai slug and SCX_AI enum key to the SCX.ai display name and logo", () => { const fromSlug = getProviderLogoAndName("scx-ai"); expect(fromSlug.displayName).toBe(Providers.SCX_AI); @@ -201,6 +212,10 @@ describe("provider_info_helpers", () => { expect(getPlaceholder(Providers.Vertex_AI)).toBe("gemini-pro"); }); + it("should return a gondola model placeholder for GONDOLA provider", () => { + expect(getPlaceholder(Providers.GONDOLA)).toBe("gondola/deepseek-v3.2"); + }); + it("should return an scx-ai model placeholder for SCX_AI provider", () => { expect(getPlaceholder(Providers.SCX_AI)).toBe("scx-ai/GLM-5.2"); }); diff --git a/ui/litellm-dashboard/src/components/provider_info_helpers.tsx b/ui/litellm-dashboard/src/components/provider_info_helpers.tsx index de83cd00790..c892c22c4b1 100644 --- a/ui/litellm-dashboard/src/components/provider_info_helpers.tsx +++ b/ui/litellm-dashboard/src/components/provider_info_helpers.tsx @@ -22,6 +22,7 @@ import fireworksLogo from "../../public/assets/logos/fireworks.svg"; import friendliLogo from "../../public/assets/logos/friendli.svg"; import gigachatLogo from "../../public/assets/logos/gigachat.svg"; import githubCopilotLogo from "../../public/assets/logos/github_copilot.svg"; +import gondolaLogo from "../../public/assets/logos/gondola.svg"; import googleLogo from "../../public/assets/logos/google.svg"; import groqLogo from "../../public/assets/logos/groq.svg"; import huggingfaceLogo from "../../public/assets/logos/huggingface.svg"; @@ -111,6 +112,7 @@ export enum Providers { GALADRIEL = "Galadriel", GIGACHAT = "GigaChat", GITHUB_COPILOT = "Github Copilot", + GONDOLA = "Gondola", Google_AI_Studio = "Google AI Studio", GradientAI = "GradientAI", Groq = "Groq", @@ -226,6 +228,7 @@ export const provider_map: Record = { GALADRIEL: "galadriel", GIGACHAT: "gigachat", GITHUB_COPILOT: "github_copilot", + GONDOLA: "gondola", Google_AI_Studio: "gemini", GradientAI: "gradient_ai", Groq: "groq", @@ -335,6 +338,7 @@ export const providerLogoMap: Partial> = { [Providers.FRIENDLIAI]: friendliLogo.src, [Providers.GIGACHAT]: gigachatLogo.src, [Providers.GITHUB_COPILOT]: githubCopilotLogo.src, + [Providers.GONDOLA]: gondolaLogo.src, [Providers.Google_AI_Studio]: googleLogo.src, [Providers.Groq]: groqLogo.src, [Providers.Hosted_Vllm]: vllmLogo.src, @@ -433,6 +437,7 @@ const providerPlaceholderMap: Partial> = { [Providers.Cursor]: "cursor/claude-4-sonnet", [Providers.DeepInfra]: "deepinfra/", [Providers.FalAI]: "fal_ai/fal-ai/flux-pro/v1.1-ultra", + [Providers.GONDOLA]: "gondola/deepseek-v3.2", [Providers.Google_AI_Studio]: "gemini-pro", [Providers.JinaAI]: "jina_ai/", [Providers.NVIDIA_RIVA]: "nvidia_riva/nvidia/parakeet-ctc-1_1b-asr",