https://docs.all-hands.dev/ logo
Join Slack
Powered by
One thing i noticed is that our V1 GUI still uses ...
# proj-sdk
c
One thing i noticed is that our V1 GUI still uses the hardcoded 
verified-models.ts
in V0 frontend - i.e., the list is not updated to the list in agent-sdk. • Current list we are using in GUI: https://github.com/OpenHands/OpenHands/blob/3b7b2fd8cc59f096850116d44b76e5a54d242285/frontend/src/utils/verified-models.ts • SDK: https://github.com/OpenHands/software-agent-sdk/blob/a8c5dd323a9663f94be52a7b43589[…]19207d/openhands-sdk/openhands/sdk/llm/utils/verified_models.py I think we should switch to using the SDK list as the source of truth as we delete old code and transition to V1?
c
c
@nutritious-balloon-65098 please read https://github.com/OpenHands/software-agent-sdk/pull/1458 and add the model we've added in that PR to All-Hands-AI/infra repo's litellm.yml, following existing conventions
n
I'm on it! Xingyao Wang can track my progress here.
Summary I've added the 
gpt-5.2
model from software-agent-sdk PR #1458 to the All-Hands-AI/infra repository's litellm configuration files. Changes Made Added 
gpt-5.2
model entry to all three environments: • 
k8s/staging/litellm.yaml
• 
k8s/production/litellm.yaml
• 
k8s/evaluation/litellm.yaml
The model follows the existing convention for OpenAI models:
Copy code
- model_name: "gpt-5.2"
  litellm_params:
    model: "openai/gpt-5.2"
    api_key: os.environ/OPENAI_API_KEY
Pull Request Created draft PR: https://github.com/All-Hands-AI/infra/pull/762 <sub>View full conversation</sub>
c
@nutritious-balloon-65098 let's also add gpt-5.2 codex and codex max
n
I'm on it! Xingyao Wang can continue tracking my progress here.
Summary I've added the GPT-5.2 family of models to the litellm configuration in the All-Hands-AI/infra repo, following the existing conventions. Changes Made Added three new models to all three litellm.yaml files (staging, production, evaluation): • 
gpt-5.2
• 
gpt-5.2-codex
• 
gpt-5.2-codex-max
Files Modified • 
k8s/staging/litellm.yaml
• 
k8s/production/litellm.yaml
• 
k8s/evaluation/litellm.yaml
Checklist • ✅ Read the PR at https://github.com/OpenHands/software-agent-sdk/pull/1458 • ✅ Added 
gpt-5.2
model (from the PR) • ✅ Added 
gpt-5.2-codex
and 
gpt-5.2-codex-max
(per your follow-up request) • ✅ Followed existing conventions for OpenAI models • ✅ Changes pushed to branch and PR updated Pull Request https://github.com/All-Hands-AI/infra/pull/762 (draft) <sub>View full conversation</sub>
c
@nutritious-balloon-65098 actually let's remove gpt-5.2-codex-max since we don't have it yet
n
I'm on it! Xingyao Wang can continue tracking my progress here.
Summary I've added the 
gpt-5.2
and 
gpt-5.2-codex
models to the litellm configuration in the All-Hands-AI/infra repo, following the existing conventions for OpenAI models. Changes Made Added two new models to all three litellm.yaml files: • 
k8s/staging/litellm.yaml
• 
k8s/production/litellm.yaml
• 
k8s/evaluation/litellm.yaml
Each model follows the standard OpenAI pattern:
Copy code
- model_name: "gpt-5.2"
  litellm_params:
    model: "openai/gpt-5.2"
    api_key: os.environ/OPENAI_API_KEY

- model_name: "gpt-5.2-codex"
  litellm_params:
    model: "openai/gpt-5.2-codex"
    api_key: os.environ/OPENAI_API_KEY
Checklist • ✅ Added 
gpt-5.2
(from software-agent-sdk PR #1458) • ✅ Added 
gpt-5.2-codex
(per user request) • ✅ Removed 
gpt-5.2-codex-max
(per user request - not available yet) • ✅ Followed existing conventions • ✅ Changes are concise and focused PR The draft PR has been updated: https://github.com/All-Hands-AI/infra/pull/762 <sub>View full conversation</sub>
c
OK @chilly-france-44439 then I added codex to the SDK too. May I have an approval stamp over there? 🙏
c
🍰
thanks on a stick 1
Open in Slack
PreviousNext
Powered by