feat(providers): add Eden AI as an OpenAI-compatible provider

core/bifrost.go

@@ -26,6 +26,7 @@ import (
 	"github.com/maximhq/bifrost/core/providers/bedrock"
 	"github.com/maximhq/bifrost/core/providers/cerebras"
 	"github.com/maximhq/bifrost/core/providers/cohere"
+	"github.com/maximhq/bifrost/core/providers/edenai"
 	"github.com/maximhq/bifrost/core/providers/elevenlabs"
 	"github.com/maximhq/bifrost/core/providers/fireworks"
 	"github.com/maximhq/bifrost/core/providers/gemini"
@@ -4056,6 +4057,8 @@ func (bifrost *Bifrost) createBaseProvider(providerKey schemas.ModelProvider, co
 		return runware.NewRunwareProvider(config, bifrost.logger)
 	case schemas.Fireworks:
 		return fireworks.NewFireworksProvider(config, bifrost.logger)
+	case schemas.EdenAI:
+		return edenai.NewEdenAIProvider(config, bifrost.logger)
 	default:
 		return nil, fmt.Errorf("unsupported provider: %s", targetProviderKey)
 	}

core/providers/edenai/edenai_live_test.go

@@ -0,0 +1,52 @@
+package edenai_test
+
+import (
+	"context"
+	"os"
+	"testing"
+
+	bifrost "github.com/maximhq/bifrost/core"
+	"github.com/maximhq/bifrost/core/providers/edenai"
+	"github.com/maximhq/bifrost/core/schemas"
+)
+
+// TestEdenAILiveChatCompletion performs a real chat completion against the Eden AI
+// API through the Bifrost provider. It is skipped unless EDENAI_API_KEY is set,
+// so CI stays hermetic.
+func TestEdenAILiveChatCompletion(t *testing.T) {
+	apiKey := os.Getenv("EDENAI_API_KEY")
+	if apiKey == "" {
+		t.Skip("EDENAI_API_KEY not set; skipping live Eden AI test")
+	}
+
+	config := &schemas.ProviderConfig{}
+	config.CheckAndSetDefaults()
+	provider, err := edenai.NewEdenAIProvider(config, bifrost.NewNoOpLogger())
+	if err != nil {
+		t.Fatalf("NewEdenAIProvider failed: %v", err)
+	}
+
+	ctx := schemas.NewBifrostContext(context.Background(), schemas.NoDeadline)
+	key := schemas.Key{Value: *schemas.NewSecretVar(apiKey), Models: schemas.WhiteList{"*"}}
+	content := "Reply with exactly: BIFROST-EDEN-OK"
+	request := &schemas.BifrostChatRequest{
+		Provider: schemas.EdenAI,
+		Model:    "openai/gpt-4o-mini",
+		Input: []schemas.ChatMessage{
+			{Role: schemas.ChatMessageRoleUser, Content: &schemas.ChatMessageContent{ContentStr: &content}},
+		},
+	}
+
+	resp, bifErr := provider.ChatCompletion(ctx, key, request)
+	if bifErr != nil {
+		msg := ""
+		if bifErr.Error != nil {
+			msg = bifErr.Error.Message
+		}
+		t.Fatalf("live ChatCompletion returned error: %s", msg)
+	}
+	if resp == nil || len(resp.Choices) == 0 {
+		t.Fatalf("expected at least one choice in the response, got %+v", resp)
+	}
+	t.Logf("Eden AI live chat completion succeeded via Bifrost provider (%d choice(s))", len(resp.Choices))
+}

core/providers/edenai/edenai_test.go

@@ -0,0 +1,43 @@
+package edenai
+
+import (
+	"testing"
+
+	"github.com/maximhq/bifrost/core/schemas"
+)
+
+// Compile-time check that edenaiProvider satisfies the full Provider interface.
+var _ schemas.Provider = (*edenaiProvider)(nil)
+
+func TestEdenAIProviderConstructorDefaults(t *testing.T) {
+	t.Parallel()
+
+	config := &schemas.ProviderConfig{}
+	config.CheckAndSetDefaults()
+	provider, err := NewEdenAIProvider(config, nil)
+	if err != nil {
+		t.Fatalf("NewEdenAIProvider failed: %v", err)
+	}
+	if provider.GetProviderKey() != schemas.EdenAI {
+		t.Errorf("expected provider key %s, got %s", schemas.EdenAI, provider.GetProviderKey())
+	}
+	if provider.networkConfig.BaseURL != "https://api.edenai.run/v3" {
+		t.Errorf("expected base URL https://api.edenai.run/v3, got %s", provider.networkConfig.BaseURL)
+	}
+}
+
+func TestEdenAIUnsupportedOperations(t *testing.T) {
+	t.Parallel()
+
+	p := &edenaiProvider{}
+
+	if _, err := p.TextCompletion(nil, schemas.Key{}, nil); err == nil {
+		t.Error("expected TextCompletion to be unsupported")
+	}
+	if _, err := p.Embedding(nil, schemas.Key{}, nil); err == nil {
+		t.Error("expected Embedding to be unsupported")
+	}
+	if _, err := p.Speech(nil, schemas.Key{}, nil); err == nil {
+		t.Error("expected Speech to be unsupported")
+	}
+}

core/schemas/bifrost.go

@@ -67,6 +67,7 @@ const (
 	Runway      ModelProvider = "runway"
 	Runware     ModelProvider = "runware"
 	Fireworks   ModelProvider = "fireworks"
+	EdenAI      ModelProvider = "edenai"
 )
 
 // SupportedBaseProviders is the list of base providers allowed for custom providers.
@@ -108,6 +109,7 @@ var StandardProviders = []ModelProvider{
 	Runway,
 	Runware,
 	Fireworks,
+	EdenAI,
 }
 
 // RequestType represents the type of request being made to a provider.

docs/docs.json

@@ -136,6 +136,7 @@
                   "providers/supported-providers/cerebras",
                   "providers/supported-providers/cohere",
                   "providers/supported-providers/databricks",
+                  "providers/supported-providers/edenai",
                   "providers/supported-providers/elevenlabs",
                   "providers/supported-providers/fireworks",
                   "providers/supported-providers/gemini",

docs/providers/supported-providers/edenai.mdx

@@ -0,0 +1,85 @@
+---
+title: "Eden AI"
+description: "Eden AI integration guide - a European, OpenAI-compatible gateway to 100+ models with a single key."
+icon: "globe"
+---
+
+## Overview
+
+[Eden AI](https://www.edenai.co/) is a **European, OpenAI-compatible** gateway that provides access to 100+ models from many providers (Mistral, OpenAI, Anthropic, Google, Cohere, DeepSeek, and more) through a single unified API and one key. Bifrost delegates to the OpenAI implementation, pointing at Eden's endpoint (`https://api.edenai.run/v3`). Key points:
+
+- **One key, many providers** — route to models from many vendors through a single `EDENAI_API_KEY`.
+- **EU-based gateway** — requests are served through an Europe-based gateway with GDPR-aligned data handling, useful for data-residency requirements.
+- **OpenAI-compatible** — chat completions, streaming (SSE), tool calling, and `reasoning_effort` are handled through the OpenAI-compatible surface.
+- **Model naming** — models keep the underlying `provider/model` form, e.g. `mistral/mistral-small-latest`, `openai/gpt-4o-mini`, `anthropic/claude-sonnet-4-5`.
+
+### Supported Operations
+
+| Operation | Non-Streaming | Streaming | Endpoint |
+|-----------|---------------|-----------|----------|
+| Chat Completions | ✅ | ✅ | `/chat/completions` |
+| Responses API | ✅ | ✅ | (delegated to Chat Completions) |
+| List Models | ✅ | - | `/models` |
+| Text Completions | ❌ | ❌ | - |
+| Embeddings | ❌ | ❌ | - |
+| Image Generation | ❌ | ❌ | - |
+| Speech (TTS) | ❌ | ❌ | - |
+| Transcriptions (STT) | ❌ | ❌ | - |
+
+<Note>
+Eden AI exposes its API under the `/v3` base path (`https://api.edenai.run/v3`), so the chat completions endpoint is `https://api.edenai.run/v3/chat/completions`. Errors are returned in the standard OpenAI error envelope.
+</Note>
+
+## Setup & Configuration
+
+Configure Eden AI as a provider.
+
+<Tabs>
+<Tab title="Web UI">
+
+1. Navigate to **Models** > **Model Providers**. If **Eden AI** is not listed under **Configured Providers**, click **Add New Provider** and select **Eden AI**.
+2. Click **Add Key** or edit an existing key.
+3. Set a name for your key.
+4. Paste your API key directly or use an environment variable (for example, `env.EDENAI_API_KEY`).
+5. Set **Allowed Models** to **All Models** (default) or a specific allowlist.
+6. Save the provider configuration.
+
+</Tab>
+<Tab title="config.json">
+
+```json
+{
+  "providers": {
+    "edenai": {
+      "keys": [
+        {
+          "name": "edenai-key-1",
+          "value": "env.EDENAI_API_KEY",
+          "models": [
+            "*"
+          ],
+          "weight": 1.0
+        }
+      ]
+    }
+  }
+}
+```
+
+The default base URL is `https://api.edenai.run/v3`. You can override it via the provider's `network_config.base_url` if needed.
+
+</Tab>
+</Tabs>
+
+## Usage
+
+```bash
+curl -X POST http://localhost:8080/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model": "edenai/mistral/mistral-small-latest",
+    "messages": [{"role": "user", "content": "Hello from a sovereign EU stack!"}]
+  }'
+```
+
+Get an API key from your [Eden AI account](https://www.edenai.co/) and see the [Eden AI documentation](https://docs.edenai.co/) for the full list of available models.

docs/providers/supported-providers/overview.mdx

@@ -31,6 +31,7 @@ The following table summarizes which operations are supported by each provider v
 | Ollama (`ollama/<model>`)            | ✅     | ✅   | ✅            | ✅   | ✅            | ✅        | ✅                 | ❌     | ❌              | ❌         | ❌                  | ❌              | ✅         | ❌  | ❌           | ❌  | ❌           | ❌    | ❌    | ❌           | ❌     | ❌  | ❌    | ❌          | ❌         | ❌          | ❌                   |
 | OpenAI (`openai/<model>`)            | ✅     | ✅   | ✅            | ✅   | ✅            | ✅        | ✅                 | ✅     | ✅              | ✅         | ✅                  | ✅              | ✅         | ✅  | ✅           | ✅  | ✅           | ✅    | ✅    | ✅           | ❌     | ❌  | ✅    | ✅          | ✅         | ✅          | ✅                   |
 | OpenRouter (`openrouter/<model>`)    | ✅     | ✅   | ✅            | ✅   | ✅            | ✅        | ✅                 | ❌     | ❌              | ❌         | ❌                  | ❌              | ✅         | ❌  | ❌           | ❌  | ❌           | ❌    | ❌    | ❌           | ❌     | ❌  | ❌    | ❌          | ❌         | ❌          | ❌                   |
+| Eden AI (`edenai/<model>`)           | ✅     | ❌   | ❌            | ✅   | ✅            | ✅        | ✅                 | ❌     | ❌              | ❌         | ❌                  | ❌              | ❌         | ❌  | ❌           | ❌  | ❌           | ❌    | ❌    | ❌           | ❌     | ❌  | ❌    | ❌          | ❌         | ❌          | ❌                   |
 | Parasail (`parasail/<model>`)        | ✅     | ❌   | ❌            | ✅   | ✅            | ✅        | ✅                 | ❌     | ❌              | ❌         | ❌                  | ❌              | ❌         | ❌  | ❌           | ❌  | ❌           | ❌    | ❌    | ❌           | ❌     | ❌  | ❌    | ❌          | ❌         | ❌          | ❌                   |
 | Perplexity (`perplexity/<model>`)    | ❌     | ❌   | ❌            | ✅   | ✅            | ✅        | ✅                 | ❌     | ❌              | ❌         | ❌                  | ❌              | ❌         | ❌  | ❌           | ❌  | ❌           | ❌    | ❌    | ❌           | ❌     | ❌  | ❌    | ❌          | ❌         | ❌          | ❌                   |
 | Replicate (`replicate/<model>`)      | ✅     | ✅   | ✅            | ✅   | ✅            | ✅        | ✅                 | ✅     | ✅              | ✅         | ✅                  | ❌              | ❌         | ❌  | ❌           | ❌  | ❌           | ✅    | ❌    | ❌           | ❌     | ❌  | ✅    | ❌          | ❌         | ❌          | ❌                   |

transports/config.schema.json

@@ -335,6 +335,9 @@
         "openrouter": {
           "$ref": "#/$defs/provider"
         },
+        "edenai": {
+          "$ref": "#/$defs/provider"
+        },
         "sgl": {
           "$ref": "#/$defs/provider_with_sgl_config"
         },
@@ -5459,7 +5462,8 @@
             "vllm",
             "runway",
             "runware",
-            "fireworks"
+            "fireworks",
+            "edenai"
           ],
           "description": "Base provider type to extend"
         },