Add chutes provider (#42)

Stenn Kool created 2 months ago

* feat: add Chutes AI provider with Qwen, GLM and Kimi models

* feat: update model pricing and add new DeepSeek V3.1 and Qwen3 models

Change summary

internal/providers/configs/chutes.json | 242 ++++++++++++++++++++++++++++
internal/providers/providers.go        |   8 
pkg/catwalk/provider.go                |   2 
3 files changed, 252 insertions(+)

Detailed changes

internal/providers/configs/chutes.json

@@ -0,0 +1,242 @@
+{
+  "name": "Chutes",
+  "id": "chutes",
+  "type": "openai",
+  "api_key": "$CHUTES_API_KEY",
+  "api_endpoint": "https://llm.chutes.ai/v1",
+  "default_large_model_id": "Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8",
+  "default_small_model_id": "Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8",
+  "models": [
+    {
+      "id": "Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8",
+      "name": "Qwen3 Coder 480B A35B Instruct (FP8)",
+      "cost_per_1m_in": 0.2,
+      "cost_per_1m_out": 0.8,
+      "context_window": 262000,
+      "default_max_tokens": 32768,
+      "can_reason": true,
+      "has_reasoning_efforts": true,
+      "default_reasoning_efforts": "medium",
+      "supports_attachments": true
+    },
+    {
+      "id": "zai-org/GLM-4.5-FP8",
+      "name": "GLM 4.5 FP8",
+      "cost_per_1m_in": 0.0,
+      "cost_per_1m_out": 0.0,
+      "context_window": 98000,
+      "default_max_tokens": 32768,
+      "can_reason": true,
+      "has_reasoning_efforts": true,
+      "default_reasoning_efforts": "medium",
+      "supports_attachments": true
+    },
+    {
+      "id": "moonshotai/Kimi-K2-Instruct-75k",
+      "name": "Kimi K2 Instruct",
+      "cost_per_1m_in": 0.15,
+      "cost_per_1m_out": 0.59,
+      "context_window": 75000,
+      "default_max_tokens": 32768,
+      "can_reason": true,
+      "has_reasoning_efforts": true,
+      "default_reasoning_efforts": "medium",
+      "supports_attachments": true
+    },
+    {
+      "id": "deepseek-ai/DeepSeek-R1-0528",
+      "name": "DeepSeek R1 0528",
+      "cost_per_1m_in": 0.18,
+      "cost_per_1m_out": 0.72,
+      "context_window": 75000,
+      "default_max_tokens": 32768,
+      "can_reason": true,
+      "has_reasoning_efforts": true,
+      "default_reasoning_efforts": "medium",
+      "supports_attachments": true
+    },
+    {
+      "id": "deepseek-ai/DeepSeek-R1-0528-Qwen3-8B",
+      "name": "DeepSeek R1 0528 Qwen3 8B",
+      "cost_per_1m_in": 0.02,
+      "cost_per_1m_out": 0.07,
+      "context_window": 32768,
+      "default_max_tokens": 8192,
+      "can_reason": false,
+      "has_reasoning_efforts": false,
+      "supports_attachments": true
+    },
+    {
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Llama-70B",
+      "name": "DeepSeek R1 Distill Llama 70B",
+      "cost_per_1m_in": 0.03,
+      "cost_per_1m_out": 0.14,
+      "context_window": 65536,
+      "default_max_tokens": 8192,
+      "can_reason": false,
+      "has_reasoning_efforts": false,
+      "supports_attachments": true
+    },
+    {
+      "id": "tngtech/DeepSeek-R1T-Chimera",
+      "name": "DeepSeek R1T Chimera",
+      "cost_per_1m_in": 0.18,
+      "cost_per_1m_out": 0.72,
+      "context_window": 131072,
+      "default_max_tokens": 32768,
+      "can_reason": true,
+      "has_reasoning_efforts": true,
+      "default_reasoning_efforts": "medium",
+      "supports_attachments": true
+    },
+    {
+      "id": "tngtech/DeepSeek-TNG-R1T2-Chimera",
+      "name": "DeepSeek TNG R1T2 Chimera",
+      "cost_per_1m_in": 0.20,
+      "cost_per_1m_out": 0.80,
+      "context_window": 262144,
+      "default_max_tokens": 65536,
+      "can_reason": true,
+      "has_reasoning_efforts": true,
+      "default_reasoning_efforts": "high",
+      "supports_attachments": true
+    },
+    {
+      "id": "deepseek-ai/DeepSeek-V3-0324",
+      "name": "DeepSeek V3 0324",
+      "cost_per_1m_in": 0.18,
+      "cost_per_1m_out": 0.72,
+      "context_window": 75000,
+      "default_max_tokens": 32768,
+      "can_reason": true,
+      "has_reasoning_efforts": true,
+      "default_reasoning_efforts": "medium",
+      "supports_attachments": true
+    },
+    {
+      "id": "chutesai/Devstral-Small-2505",
+      "name": "Devstral Small 2505",
+      "cost_per_1m_in": 0.02,
+      "cost_per_1m_out": 0.08,
+      "context_window": 32768,
+      "default_max_tokens": 8192,
+      "can_reason": false,
+      "has_reasoning_efforts": false,
+      "supports_attachments": true
+    },
+    {
+      "id": "zai-org/GLM-4.5-Air",
+      "name": "GLM 4.5 Air",
+      "cost_per_1m_in": 0.0,
+      "cost_per_1m_out": 0.0,
+      "context_window": 131072,
+      "default_max_tokens": 32768,
+      "can_reason": true,
+      "has_reasoning_efforts": true,
+      "default_reasoning_efforts": "medium",
+      "supports_attachments": true
+    },
+    {
+      "id": "openai/gpt-oss-120b",
+      "name": "GPT OSS 120B",
+      "cost_per_1m_in": 0.10,
+      "cost_per_1m_out": 0.41,
+      "context_window": 131072,
+      "default_max_tokens": 32768,
+      "can_reason": true,
+      "has_reasoning_efforts": true,
+      "default_reasoning_efforts": "medium",
+      "supports_attachments": true
+    },
+    {
+      "id": "chutesai/Mistral-Small-3.2-24B-Instruct-2506",
+      "name": "Mistral Small 3.2 24B Instruct 2506",
+      "cost_per_1m_in": 0.02,
+      "cost_per_1m_out": 0.08,
+      "context_window": 32768,
+      "default_max_tokens": 8192,
+      "can_reason": false,
+      "has_reasoning_efforts": false,
+      "supports_attachments": true
+    },
+    {
+      "id": "Qwen/Qwen3-235B-A22B-Instruct-2507",
+      "name": "Qwen3 235B A22B Instruct 2507",
+      "cost_per_1m_in": 0.08,
+      "cost_per_1m_out": 0.31,
+      "context_window": 32768,
+      "default_max_tokens": 8192,
+      "can_reason": false,
+      "has_reasoning_efforts": false,
+      "supports_attachments": true
+    },
+    {
+      "id": "Qwen/Qwen3-30B-A3B",
+      "name": "Qwen3 30B A3B",
+      "cost_per_1m_in": 0.02,
+      "cost_per_1m_out": 0.08,
+      "context_window": 32768,
+      "default_max_tokens": 8192,
+      "can_reason": false,
+      "has_reasoning_efforts": false,
+      "supports_attachments": true
+    },
+    {
+      "id": "Qwen/Qwen3-235B-A22B-Thinking-2507",
+      "name": "Qwen3 235B A22B Thinking 2507",
+      "cost_per_1m_in": 0.08,
+      "cost_per_1m_out": 0.31,
+      "context_window": 32768,
+      "default_max_tokens": 8192,
+      "can_reason": true,
+      "has_reasoning_efforts": true,
+      "default_reasoning_efforts": "high",
+      "supports_attachments": true
+    },
+    {
+      "id": "deepseek-ai/DeepSeek-V3.1",
+      "name": "DeepSeek V3.1",
+      "cost_per_1m_in": 0.20,
+      "cost_per_1m_out": 0.80,
+      "context_window": 163840,
+      "default_max_tokens": 32768,
+      "can_reason": false,
+      "has_reasoning_efforts": false,
+      "supports_attachments": true
+    },
+    {
+      "id": "deepseek-ai/DeepSeek-V3.1:THINKING",
+      "name": "DeepSeek V3.1 Reasoning",
+      "cost_per_1m_in": 0.20,
+      "cost_per_1m_out": 0.80,
+      "context_window": 163840,
+      "default_max_tokens": 32768,
+      "can_reason": true,
+      "has_reasoning_efforts": true,
+      "default_reasoning_efforts": "medium",
+      "supports_attachments": true
+    },
+    {
+      "id": "Qwen/Qwen3-30B-A3B-Instruct-2507",
+      "name": "Qwen3 30B A3B Instruct 2507",
+      "cost_per_1m_in": 0.05,
+      "cost_per_1m_out": 0.20,
+      "context_window": 262144,
+      "default_max_tokens": 32768,
+      "can_reason": false,
+      "has_reasoning_efforts": false,
+      "supports_attachments": true
+    },
+    {
+      "id": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
+      "name": "Qwen3 Coder 30B A3B Instruct",
+      "cost_per_1m_in": 0.00,
+      "cost_per_1m_out": 0.00,
+      "context_window": 262144,
+      "default_max_tokens": 32768,
+      "can_reason": false,
+      "has_reasoning_efforts": false,
+      "supports_attachments": true
+    }
+  ]
+}

internal/providers/providers.go

@@ -48,6 +48,9 @@ var cerebrasConfig []byte
 //go:embed configs/venice.json
 var veniceConfig []byte
 
+//go:embed configs/chutes.json
+var chutesConfig []byte
+
 //go:embed configs/deepseek.json
 var deepSeekConfig []byte
 
@@ -68,6 +71,7 @@ var providerRegistry = []ProviderFunc{
 	lambdaProvider,
 	cerebrasProvider,
 	veniceProvider,
+	chutesProvider,
 	deepSeekProvider,
 }
 
@@ -141,6 +145,10 @@ func veniceProvider() catwalk.Provider {
 	return loadProviderFromConfig(veniceConfig)
 }
 
+func chutesProvider() catwalk.Provider {
+	return loadProviderFromConfig(chutesConfig)
+}
+
 func deepSeekProvider() catwalk.Provider {
 	return loadProviderFromConfig(deepSeekConfig)
 }

pkg/catwalk/provider.go

@@ -31,6 +31,7 @@ const (
 	InferenceProviderLambda     InferenceProvider = "lambda"
 	InferenceProviderCerebras   InferenceProvider = "cerebras"
 	InferenceProviderVenice     InferenceProvider = "venice"
+	InferenceProviderChutes     InferenceProvider = "chutes"
 )
 
 // Provider represents an AI provider configuration.
@@ -78,5 +79,6 @@ func KnownProviders() []InferenceProvider {
 		InferenceProviderLambda,
 		InferenceProviderCerebras,
 		InferenceProviderVenice,
+		InferenceProviderChutes,
 	}
 }