Merge pull request #1820 from cmodi-meta/llama-api-support

Add Llama API Support

Author: better629

config/examples/llamaapi-Llama-3.3-70B-Instruct.yaml

@@ -0,0 +1,6 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates
+llm:
+  api_type: llama_api
+  base_url: "https://api.llama.com/compat/v1/"
+  api_key: "YOUR_API_KEY"
+  model: "Llama-3.3-70B-Instruct"

config/examples/llamaapi-Llama-3.3-8B-Instruct.yaml

@@ -0,0 +1,6 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates
+llm:
+  api_type: llama_api
+  base_url: "https://api.llama.com/compat/v1/"
+  api_key: "YOUR_API_KEY"
+  model: "Llama-3.3-8B-Instruct"

config/examples/llamaapi-Llama-4-Scout-17B-16E-Instruct-FP8.yaml

@@ -0,0 +1,6 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates
+llm:
+  api_type: llama_api
+  base_url: "https://api.llama.com/compat/v1/"
+  api_key: "YOUR_API_KEY"
+  model: "Llama-4-Scout-17B-16E-Instruct-FP8"

config/examples/llamaapi-llama-4-Maverick-17B-128E-Instruct-FP8.yaml

@@ -0,0 +1,6 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates
+llm:
+  api_type: llama_api
+  base_url: "https://api.llama.com/compat/v1/"
+  api_key: "YOUR_API_KEY"
+  model: "Llama-4-Maverick-17B-128E-Instruct-FP8"

metagpt/configs/llm_config.py

@@ -43,6 +43,7 @@ class LLMType(Enum):
     OPENROUTER_REASONING = "openrouter_reasoning"
     BEDROCK = "bedrock"
     ARK = "ark"  # https://www.volcengine.com/docs/82379/1263482#python-sdk
+    LLAMA_API = "llama_api"
 
     def __missing__(self, key):
         return self.OPENAI

metagpt/provider/openai_api.py

@@ -52,6 +52,7 @@
         LLMType.DEEPSEEK,
         LLMType.SILICONFLOW,
         LLMType.OPENROUTER,
+        LLMType.LLAMA_API,
     ]
 )
 class OpenAILLM(BaseLLM):

metagpt/utils/token_counter.py

@@ -113,6 +113,10 @@
     "doubao-pro-128k-240515": {"prompt": 0.0007, "completion": 0.0013},
     "llama3-70b-llama3-70b-instruct": {"prompt": 0.0, "completion": 0.0},
     "llama3-8b-llama3-8b-instruct": {"prompt": 0.0, "completion": 0.0},
+    "llama-4-Scout-17B-16E-Instruct-FP8" : {"prompt": 0.0, "completion": 0.0}, # start, for Llama API
+    "llama-4-Maverick-17B-128E-Instruct-FP8": {"prompt": 0.0, "completion": 0.0},
+    "llama-3.3-8B-Instruct": {"prompt": 0.0, "completion": 0.0},
+    "llama-3.3-70B-Instruct": {"prompt": 0.0, "completion": 0.0}, # end, for Llama API
 }