Skip to content

fix(decisions): gemini-decision-1 needs a description on every choice and level - #125

Merged
VickyXAI merged 3 commits into
mainfrom
fix/gemini-decision-descriptions
Nov 18, 2026
Merged

VickyXAI merged 3 commits into
mainfrom
fix/gemini-decision-descriptions

Conversation

@VickyXAI

Copy link
Copy Markdown
Contributor

Problem

gemini-decision-1 launched on sol.blockrun.ai and blockrun.ai on 2026-11-18. It needs a description string on every choice and every score level. An empty string is fine.

Without one, the gateway answers 400 before payment. Nothing is charged, but the call fails. The README and JSDoc examples leave some descriptions out, so copying them with model: 'gemini-decision-1' fails.

Probed on sol on 2026-11-18. Each probe was refused before payment, so they cost nothing:

Request Result
Gemini, choice without description 400 questions[0].choices[0].description must be a string (it may be empty)
Gemini, score level without description 400 questions[0].levels[0].description must be a string (it may be empty)
Gemini, every item described 402 (valid)
luna, with descriptions 402 (valid)

Change

  • buildDecisionsBody now throws an "unpaid" error before sending when the model is gemini-decision-1 or google/gemini-decision-1 and a choice or level has no description string. The error names the exact item.
  • luna is unchanged.
  • The README and JSDoc examples give every choice and level a description, so they work on both models.
  • CHANGELOG: one Unreleased entry.

Tests

  • New gemini-decision-1 descriptions suite. Five of its cases failed before the fix:
    • a choice without a description, for both id forms;
    • a level without a description, for both id forms;
    • the refusal is an "unpaid" error (nothing was sent).
  • Controls:
    • every item described (empty strings accepted) passes;
    • luna (no model, bare id, openai/ id) does not require descriptions.
  • vitest run: 1202 passed, 12 skipped. tsc --noEmit and eslint src/ are clean.
  • The gateway side was live-tested in the matching Python SDK PR (fix(decisions): gemini-decision-1 needs a description on every choice and level blockrun-llm#117). There, the README example answered on gemini-decision-1 and on gpt-6-luna on sol.

Noticed, not changed here

On the same input, Gemini returns score as a decimal (1.43), while luna returns an integer index (1). The README describes score as an index. The docs or the response type may need a note. That is a separate decision.

🤖 Generated with Claude Code

1bcMax added 3 commits November 18, 2026 10:59
… and level

Gemini Decision 1 refuses a choice or score level without a description string
(gateway 400 before payment, probed on sol.blockrun.ai 2026-11-18).
buildDecisionsBody now throws an unpaid error before sending, naming the item.
gpt-6-luna is unchanged. README and JSDoc examples give every item a
description, so they work on both models.
@VickyXAI
VickyXAI merged commit 1142f78 into main Nov 18, 2026
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant