feat: support hyperbolic client

Argcfile.sh

@@ -320,6 +320,7 @@ _argc_before() {
         github,gpt-4o-mini,https://models.inference.ai.azure.com \
         groq,llama-3.1-8b-instant,https://api.groq.com/openai/v1 \
         hunyuan,hunyuan-large,https://api.hunyuan.cloud.tencent.com/v1 \
+        hyperbolic,meta-llama/Meta-Llama-3.1-8B-Instruct,https://api.hyperbolic.xyz/v1 \
         lingyiwanwu,yi-lightning,https://api.lingyiwanwu.com/v1 \
         minimax,MiniMax-Text-01,https://api.minimax.chat/v1 \
         mistral,mistral-small-latest,https://api.mistral.ai/v1 \

config.example.yaml

@@ -328,6 +328,12 @@ clients:
     api_base: https://api.together.xyz/v1
     api_key: xxx
 
+  # See https://docs.hyperbolic.xyz/docs/getting-started
+  - type: openai-compatible
+    name: hyperbolic
+    api_base: https://api.hyperbolic.xyz/v1
+    api_key: xxx
+
   # ----- RAG dedicated -----
 
   # See https://jina.ai

models.yaml

@@ -1782,6 +1782,58 @@
       max_input_tokens: 8192
       input_price: 0.1
 
+# Links:
+#  - https://app.hyperbolic.xyz/models
+#  - https://docs.hyperbolic.xyz/docs/hyperbolic-ai-inference-pricing
+- provider: hyperbolic
+  models:
+    - name: meta-llama/Llama-3.3-70B-Instruct
+      max_input_tokens: 131072
+      input_price: 0.4
+      output_price: 0.4
+    - name: meta-llama/Meta-Llama-3.1-405B-Instruct
+      max_input_tokens: 32768
+      input_price: 4
+      output_price: 4
+    - name: meta-llama/Meta-Llama-3.1-70B-Instruct
+      max_input_tokens: 32768
+      input_price: 0.4
+      output_price: 0.4
+    - name: meta-llama/Meta-Llama-3.1-8B-Instruct
+      max_input_tokens: 32768
+      input_price: 0.1
+      output_price: 0.1
+    - name: Qwen/Qwen2.5-72B-Instruct
+      max_input_tokens: 32768
+      input_price: 0.4
+      output_price: 0.4
+    - name: Qwen/Qwen2.5-Coder-32B-Instruct
+      max_input_tokens: 131072
+      input_price: 0.2
+      output_price: 0.2
+    - name: Qwen/QwQ-32B-Preview
+      max_input_tokens: 32768
+      input_price: 0.2
+      output_price: 0.2
+    - name: Qwen/Qwen2-VL-72B-Instruct
+      max_input_tokens: 32768
+      input_price: 0.4
+      output_price: 0.4
+      supports_vision: true
+    - name: deepseek-ai/DeepSeek-V3
+      max_input_tokens: 131072
+      input_price: 0.25
+      output_price: 0.25
+    - name: deepseek-ai/DeepSeek-R1
+      max_input_tokens: 163840
+      input_price: 2
+      output_price: 2
+    - name: mistralai/Pixtral-12B-2409
+      max_input_tokens: 128000
+      input_price: 0.1
+      output_price: 0.1
+      supports_vision: true
+
 # Links:
 #  - https://jina.ai/models
 #  - https://api.jina.ai/redoc

src/client/mod.rs

@@ -34,7 +34,7 @@ register_client!(
     (ernie, "ernie", ErnieConfig, ErnieClient),
 );
 
-pub const OPENAI_COMPATIBLE_PROVIDERS: [(&str, &str); 22] = [
+pub const OPENAI_COMPATIBLE_PROVIDERS: [(&str, &str); 23] = [
     ("ai21", "https://api.ai21.com/studio/v1"),
     (
         "cloudflare",
@@ -46,6 +46,7 @@ pub const OPENAI_COMPATIBLE_PROVIDERS: [(&str, &str); 22] = [
     ("github", "https://models.inference.ai.azure.com"),
     ("groq", "https://api.groq.com/openai/v1"),
     ("hunyuan", "https://api.hunyuan.cloud.tencent.com/v1"),
+    ("hyperbolic", "https://api.hyperbolic.xyz/v1"),
     ("lingyiwanwu", "https://api.lingyiwanwu.com/v1"),
     ("minimax", "https://api.minimax.chat/v1"),
     ("mistral", "https://api.mistral.ai/v1"),