brian/homeclaude
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages 1 Wiki Activity Actions

Fix response truncation, blank artifacts, and offer Opus #1

Merged
brian merged 1 commit from fix/response-truncation-and-opus-discovery into main 2026-06-19 19:48:56 +00:00
Conversation 0 Commits 1 Files changed 1 +34 -7
brian commented 2026-06-19 19:48:50 +00:00
Owner
Copy link

  • Chat streams capped max_tokens at 2048, truncating long replies.
    Use each model's true output ceiling (64K, or 128K for Opus/Fable)
    via maxOutputTokens(); both chat paths already stream, so a high
    ceiling is safe and only bills for tokens actually generated.

  • Multiple artifacts in one turn could be saved blank when a later
    tool_use was cut off before its content streamed. create_artifact
    now refuses empty content and returns an error result so the model
    regenerates it via the existing tool loop.

  • Model discovery now scans the opus tier too, so the admin page can
    offer the newest Opus the API key can access. It stays opt-in: not
    seeded in DEFAULT_MODELS and never set as default. Discovered Opus
    models get sensible fallback pricing instead of $0.

Co-Authored-By: Claude Opus 4.8 noreply@anthropic.com

- Chat streams capped max_tokens at 2048, truncating long replies. Use each model's true output ceiling (64K, or 128K for Opus/Fable) via maxOutputTokens(); both chat paths already stream, so a high ceiling is safe and only bills for tokens actually generated. - Multiple artifacts in one turn could be saved blank when a later tool_use was cut off before its content streamed. create_artifact now refuses empty content and returns an error result so the model regenerates it via the existing tool loop. - Model discovery now scans the opus tier too, so the admin page can offer the newest Opus the API key can access. It stays opt-in: not seeded in DEFAULT_MODELS and never set as default. Discovered Opus models get sensible fallback pricing instead of $0. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Chat streams capped max_tokens at 2048, truncating long replies.
  Use each model's true output ceiling (64K, or 128K for Opus/Fable)
  via maxOutputTokens(); both chat paths already stream, so a high
  ceiling is safe and only bills for tokens actually generated.

- Multiple artifacts in one turn could be saved blank when a later
  tool_use was cut off before its content streamed. create_artifact
  now refuses empty content and returns an error result so the model
  regenerates it via the existing tool loop.

- Model discovery now scans the opus tier too, so the admin page can
  offer the newest Opus the API key can access. It stays opt-in: not
  seeded in DEFAULT_MODELS and never set as default. Discovered Opus
  models get sensible fallback pricing instead of $0.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
brian merged commit 2843de61bf into main 2026-06-19 19:48:56 +00:00
brian deleted branch fix/response-truncation-and-opus-discovery 2026-06-19 19:48:56 +00:00
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
brian/homeclaude!1
No description provided.