Alex 9a69e89096 fix: assistant input content must be output_text, not input_text ...

Live /v1/responses rejects input_text for the assistant role
(input[N].content[0] -> "Supported values are: 'output_text' and
'refusal'"). Revert the previous review-driven change: encode assistant
turns as output_text and user/system turns as input_text. The SDK type
union suggested input_* only, but the API enforces output_text per-role —
confirmed against the live endpoint.

2026-06-03 15:00:14 +01:00

..

handlers

fix: backup model response parsing

2026-06-01 11:30:37 +02:00

providers

feat: add OpenAI Responses API support via per-model api_flavor

2026-06-03 13:43:04 +01:00

__init__.py

testings

2023-09-26 10:03:22 +01:00

anthropic.py

feat: better logging

2026-04-28 00:14:43 +01:00

base.py

fix: backup model response parsing

2026-06-01 11:30:37 +02:00

docsgpt_provider.py

feat: better logging

2026-04-28 00:14:43 +01:00

google_ai.py

Deepseek (#2499)

2026-05-25 23:49:25 +01:00

groq.py

feat: better logging

2026-04-28 00:14:43 +01:00

llama_cpp.py

feat: better logging

2026-04-28 00:14:43 +01:00

llm_creator.py

feat: add model id on token usage (#2507)

2026-05-28 19:33:01 +01:00

novita.py

feat: better logging

2026-04-28 00:14:43 +01:00

open_router.py

feat: better logging

2026-04-28 00:14:43 +01:00

openai.py

fix: assistant input content must be output_text, not input_text

2026-06-03 15:00:14 +01:00

premai.py

feat: better logging

2026-04-28 00:14:43 +01:00

sagemaker.py

feat: better logging

2026-04-28 00:14:43 +01:00