client.lemur.summary(), .questionAnswer(), .actionItems() and .task() all call /lemur/v3/generate/*, and all 4 answer 404 Not found with a bare body. getResponse() on /lemur/v3/{id} does the same.
To rule out an account problem: with the same key, GET /v2/transcript answers 200, and GET /v2/no-such-thing-here answers 404 Not found, byte for byte what the LeMUR paths answer. So the route is gone rather than closed to me.
The Python SDK carries no LeMUR at all and this one carries all of it, so the 2 official SDKs disagree with each other.
The work moved to the LLM Gateway at llm-gateway.assemblyai.com/v1/chat/completions. One difference reaches the caller: LeMUR took transcript_ids and fetched the text itself, the gateway takes text, so fetching and capping each transcript has to happen on the way in.
I hit this porting my own client and can send a PR if you want one. Which way to go is yours to pick: drop the methods, or keep the names and point them at the gateway.
client.lemur.summary(),.questionAnswer(),.actionItems()and.task()all call/lemur/v3/generate/*, and all 4 answer404 Not foundwith a bare body.getResponse()on/lemur/v3/{id}does the same.To rule out an account problem: with the same key,
GET /v2/transcriptanswers 200, andGET /v2/no-such-thing-hereanswers404 Not found, byte for byte what the LeMUR paths answer. So the route is gone rather than closed to me.The Python SDK carries no LeMUR at all and this one carries all of it, so the 2 official SDKs disagree with each other.
The work moved to the LLM Gateway at
llm-gateway.assemblyai.com/v1/chat/completions. One difference reaches the caller: LeMUR tooktranscript_idsand fetched the text itself, the gateway takes text, so fetching and capping each transcript has to happen on the way in.I hit this porting my own client and can send a PR if you want one. Which way to go is yours to pick: drop the methods, or keep the names and point them at the gateway.