API openrouter does not work

[edisop]

I tried the documentation:

llm:
api_type: 'openrouter'
base_url: 'https://openrouter.ai/api/v1'
api_key: 'sk...'
model: meta-llama/llama-3-70b-instruct:nitro

Then I got this issue:

---

KeyError Traceback (most recent call last)
File :1

File ~/anaconda3/lib/python3.10/site-packages/nest_asyncio.py:35, in _patch_asyncio..run(main, debug)
33 task = asyncio.ensure_future(main)
34 try:
---> 35 return loop.run_until_complete(task)
36 finally:
37 if not task.done():

File ~/anaconda3/lib/python3.10/site-packages/nest_asyncio.py:90, in _patch_loop..run_until_complete(self, future)
87 if not f.done():
88 raise RuntimeError(
89 'Event loop stopped before Future completed.')
---> 90 return f.result()

File ~/anaconda3/lib/python3.10/asyncio/futures.py:201, in Future.result(self)
199 self.__log_traceback = False
200 if self._exception is not None:
--> 201 raise self._exception.with_traceback(self._exception_tb)
202 return self._result

File ~/anaconda3/lib/python3.10/asyncio/tasks.py:232, in Task.__step(failed resolving arguments)
228 try:
229 if exc is None:
230 # We use the send method directly, because coroutines
231 # don't have __iter__ and __next__ methods.
--> 232 result = coro.send(None)
233 else:
234 result = coro.throw(exc)

File ~/Desktop/AI_eval_redo/src.py:766, in main(questions, investment, n_round, add_human, output_dir)
761 logger.info(f"Processing idea: {idea}")
763 team = Team()
764 team.hire(
765 [
--> 766 RAGer(),
767 ProductManager(),
768 SimpleReviewer(),
769 Researcher(),
770 ]
771 )
773 team.invest(investment=investment)
774 team.run_project(idea)

File ~/Desktop/AI_eval_redo/src.py:488, in RAGer.init(self, **kwargs)
487 def init(self, **kwargs):
--> 488 super().init(**kwargs)
489 self._watch([ProductManage])
490 self.set_actions([RAG_TT])

[... skipping hidden 1 frame]

File ~/anaconda3/lib/python3.10/site-packages/metagpt/roles/role.py:161, in Role.validate_role_extra(self)
159 @model_validator(mode="after")
160 def validate_role_extra(self):
--> 161 self._process_role_extra()
162 return self

File ~/anaconda3/lib/python3.10/site-packages/metagpt/roles/role.py:171, in Role._process_role_extra(self)
168 self.llm = HumanProvider(None)
170 self._check_actions()
--> 171 self.llm.system_prompt = self._get_prefix()
172 self.llm.cost_manager = self.context.cost_manager
173 self._watch(kwargs.pop("watch", [UserRequirement]))

File ~/anaconda3/lib/python3.10/site-packages/metagpt/context_mixin.py:95, in ContextMixin.llm(self)
93 # print(f"class:{self.class.name}({self.name}), llm: {self._llm}, llm_config: {self._llm_config}")
94 if not self.private_llm:
---> 95 self.private_llm = self.context.llm_with_cost_manager_from_llm_config(self.config.llm)
96 return self.private_llm

File ~/anaconda3/lib/python3.10/site-packages/metagpt/context.py:107, in Context.llm_with_cost_manager_from_llm_config(self, llm_config)
105 """Return a LLM instance, fixme: support cache"""
106 # if self._llm is None:
--> 107 llm = create_llm_instance(llm_config)
108 if llm.cost_manager is None:
109 llm.cost_manager = self._select_costmanager(llm_config)

File ~/anaconda3/lib/python3.10/site-packages/metagpt/provider/llm_provider_registry.py:40, in create_llm_instance(config)
38 def create_llm_instance(config: LLMConfig) -> BaseLLM:
39 """get the default llm provider"""
---> 40 return LLM_REGISTRY.get_provider(config.api_type)(config)

File ~/anaconda3/lib/python3.10/site-packages/metagpt/provider/llm_provider_registry.py:21, in LLMProviderRegistry.get_provider(self, enum)
19 def get_provider(self, enum: LLMType):
20 """get provider instance according to the enum"""
---> 21 return self.providers[enum]

KeyError: <LLMType.OPENROUTER: 'openrouter'>

Looking at the GitHub repo, it seems like "openrouter" does not have a ".py" script under the path metagpt/provider/

Other working API like OpenAI, Anthropic all seems to have corresponding ".py" script.

[edisop]

Sorry I forgot to cite where the documentation I'm referring to that supports openrouter: https://docs.deepwisdom.ai/main/zh/guide/get_started/configuration/llm_api_configuration.html

[iorisa]

llm:
  base_url: "https://openrouter.ai/api/v1"
  api_key: "sk-or-v1-0780*****"
  model: "anthropic/claude-3-sonnet"