Merge pull request #25 from jooray/main

Kujtim Hoxha created 3 months ago

Add Venice.ai as a provider. Tested with crush.

Change summary

internal/providers/configs/venice.json | 71 ++++++++++++++++++++++++++++
internal/providers/providers.go        |  8 +++
pkg/catwalk/provider.go                |  2 
3 files changed, 81 insertions(+)

Detailed changes

internal/providers/configs/venice.json

@@ -0,0 +1,71 @@
+{
+  "name": "Venice AI",
+  "id": "venice",
+  "type": "openai",
+  "api_key": "$VENICE_API_KEY",
+  "api_endpoint": "https://api.venice.ai/api/v1",
+  "default_large_model_id": "qwen3-235b:strip_thinking_response=true",
+  "default_small_model_id": "mistral-31-24b",
+  "models": [
+    {
+      "id": "qwen3-235b:strip_thinking_response=true",
+      "name": "Venice Large (qwen3-235b)",
+      "cost_per_1m_in": 1.5,
+      "cost_per_1m_out": 6,
+      "cost_per_1m_in_cached": 0,
+      "cost_per_1m_out_cached": 0,
+      "context_window": 131072,
+      "default_max_tokens": 50000,
+      "can_reason": true,
+      "supports_attachments": false
+    },
+    {
+      "id": "qwen3-4b:strip_thinking_response=true",
+      "name": "Venice Small (qwen3-4b)",
+      "cost_per_1m_in": 0.15,
+      "cost_per_1m_out": 0.6,
+      "cost_per_1m_in_cached": 0,
+      "cost_per_1m_out_cached": 0,
+      "context_window": 32768,
+      "default_max_tokens": 25000,
+      "can_reason": true,
+      "supports_attachments": false
+    },
+    {
+      "id": "mistral-31-24b",
+      "name": "Venice Medium (mistral-31-24b)",
+      "cost_per_1m_in": 0.5,
+      "cost_per_1m_out": 2,
+      "cost_per_1m_in_cached": 0,
+      "cost_per_1m_out_cached": 0,
+      "context_window": 131072,
+      "default_max_tokens": 50000,
+      "can_reason": false,
+      "supports_attachments": true
+    },
+    {
+      "id": "llama-3.2-3b",
+      "name": "Llama 3.2 3B",
+      "cost_per_1m_in": 0.15,
+      "cost_per_1m_out": 0.6,
+      "cost_per_1m_in_cached": 0,
+      "cost_per_1m_out_cached": 0,
+      "context_window": 131072,
+      "default_max_tokens": 25000,
+      "can_reason": false,
+      "supports_attachments": false
+    },
+    {
+      "id": "llama-3.3-70b",
+      "name": "Llama 3.3 70B",
+      "cost_per_1m_in": 0.7,
+      "cost_per_1m_out": 2.8,
+      "cost_per_1m_in_cached": 0,
+      "cost_per_1m_out_cached": 0,
+      "context_window": 65536,
+      "default_max_tokens": 32000,
+      "can_reason": false,
+      "supports_attachments": false
+    }
+  ]
+}

internal/providers/providers.go

@@ -45,6 +45,9 @@ var lambdaConfig []byte
 //go:embed configs/cerebras.json
 var cerebrasConfig []byte
 
+//go:embed configs/venice.json
+var veniceConfig []byte
+
 // ProviderFunc is a function that returns a Provider.
 type ProviderFunc func() catwalk.Provider
 
@@ -61,6 +64,7 @@ var providerRegistry = []ProviderFunc{
 	openRouterProvider,
 	lambdaProvider,
 	cerebrasProvider,
+	veniceProvider,
 }
 
 // GetAll returns all registered providers.
@@ -128,3 +132,7 @@ func lambdaProvider() catwalk.Provider {
 func cerebrasProvider() catwalk.Provider {
 	return loadProviderFromConfig(cerebrasConfig)
 }
+
+func veniceProvider() catwalk.Provider {
+	return loadProviderFromConfig(veniceConfig)
+}

pkg/catwalk/provider.go

@@ -30,6 +30,7 @@ const (
 	InferenceProviderOpenRouter InferenceProvider = "openrouter"
 	InferenceProviderLambda     InferenceProvider = "lambda"
 	InferenceProviderCerebras   InferenceProvider = "cerebras"
+	InferenceProviderVenice     InferenceProvider = "venice"
 )
 
 // Provider represents an AI provider configuration.
@@ -76,5 +77,6 @@ func KnownProviders() []InferenceProvider {
 		InferenceProviderOpenRouter,
 		InferenceProviderLambda,
 		InferenceProviderCerebras,
+		InferenceProviderVenice,
 	}
 }