fix(openai): support all o series models (#1031)

Before this change, since `max_completion_tokens` was not set for `o` series models, the completion request will time out sometimes. This makes sure it converts the `max_tokens` parameter to `max_completion_tokens` for `o` series models.

I tested this change with `gpt-4o-mini`, `o1-mini` and `o3-mini`, and they all still work as expected.
This commit is contained in:
Larry Lv 2025-01-04 21:23:33 -08:00 committed by GitHub
parent 81e5f1937f
commit ec5d1abf34
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194

View File

@ -54,6 +54,10 @@ M.get_user_message = function(opts)
)
end
M.is_o_series_model = function(model)
return model and string.match(model, "^o%d+") ~= nil
end
M.parse_messages = function(opts)
local messages = {}
local provider = P[Config.provider]
@ -61,7 +65,7 @@ M.parse_messages = function(opts)
-- NOTE: Handle the case where the selected model is the `o1` model
-- "o1" models are "smart" enough to understand user prompt as a system prompt in this context
if base.model and string.find(base.model, "o1") then
if M.is_o_series_model(base.model) then
table.insert(messages, { role = "user", content = opts.system_prompt })
else
table.insert(messages, { role = "system", content = opts.system_prompt })
@ -150,9 +154,10 @@ M.parse_curl_args = function(provider, code_opts)
headers["Authorization"] = "Bearer " .. api_key
end
-- NOTE: When using "o1" set the supported parameters only
-- NOTE: When using "o" series set the supported parameters only
local stream = true
if base.model and string.find(base.model, "o1") then
if M.is_o_series_model(base.model) then
body_opts.max_completion_tokens = body_opts.max_tokens
body_opts.max_tokens = nil
body_opts.temperature = 1
end