feat: support github client

Argcfile.sh

@@ -92,6 +92,7 @@ OPENAI_COMPATIBLE_PLATFORMS=( \
   together,meta-llama/Llama-3-8b-chat-hf,https://api.together.xyz/v1 \
   zhipuai,glm-4-0520,https://open.bigmodel.cn/api/paas/v4 \
   lingyiwanwu,yi-large,https://api.lingyiwanwu.com/v1 \
+  github,meta-llama-3.1-8b-instruct,https://models.inference.ai.azure.com \
 )
 
 # @cmd Chat with any LLM api 

config.example.yaml

@@ -268,6 +268,12 @@ clients:
     api_key: xxx
     api_base: https://api.deepinfra.com/v1/openai
 
+  # See https://github.com/marketplace/models
+  - type: openai-compatible
+    name: github
+    api_key: xxx
+    api_base: https://models.inference.ai.azure.com
+
   # See https://readme.fireworks.ai/docs/quickstart
   - type: openai-compatible
     name: fireworks

models.yaml

@@ -724,6 +724,44 @@
     input_price: 0.14
     output_price: 0.14
 
+- platform: github
+  # docs:
+  #   - https://github.com/marketplace/models
+  #   - https://deepinfra.com/pricing
+  models:
+    - name: gpt-4o
+      max_input_tokens: 128000
+      supports_function_calling: true
+    - name: gpt-4o-mini
+      max_input_tokens: 128000
+      supports_function_calling: true
+    - name: meta-llama-3.1-405b-instruct
+      max_input_tokens: 128000
+    - name: meta-llama-3.1-70b-instruct
+      max_input_tokens: 128000
+    - name: meta-llama-3.1-8b-instruct
+      max_input_tokens: 128000
+    - name: meta-llama-3-70b-instruct
+      max_input_tokens: 8192
+    - name: meta-llama-3-8b-instruct
+      max_input_tokens: 8192
+    - name: mistral-large-2407
+      max_input_tokens: 128000
+      supports_function_calling: true
+    - name: mistral-nemo
+      max_input_tokens: 128000
+      supports_function_calling: true
+    - name: cohere-command-r-plus
+      max_input_tokens: 128000
+      supports_function_calling: true
+    - name: cohere-command-r
+      max_input_tokens: 128000
+      supports_function_calling: true
+    - name: phi-3-medium-128k-instruct
+      max_input_tokens: 128000
+    - name: phi-3-mini-128k-instruct
+      max_input_tokens: 128000
+
 - platform: deepinfra
   # docs:
   #   - https://deepinfra.com/models

src/client/mod.rs

@@ -40,10 +40,11 @@ register_client!(
     (qianwen, "qianwen", QianwenConfig, QianwenClient),
 );
 
-pub const OPENAI_COMPATIBLE_PLATFORMS: [(&str, &str); 14] = [
+pub const OPENAI_COMPATIBLE_PLATFORMS: [(&str, &str); 15] = [
     ("deepinfra", "https://api.deepinfra.com/v1/openai"),
     ("deepseek", "https://api.deepseek.com"),
     ("fireworks", "https://api.fireworks.ai/inference/v1"),
+    ("github", "https://models.inference.ai.azure.com"),
     ("groq", "https://api.groq.com/openai/v1"),
     ("jina", "https://api.jina.ai/v1"),
     ("lingyiwanwu", "https://api.lingyiwanwu.com/v1"),