Repository navigation
Conversation
gemini-4-neon supports thinkingLevel like gemini-4-hydrogen. Add it to the capability table and the level mapping, so minimal becomes low and low, medium, high, xhigh, max pass through on cross-protocol requests. Closes QuantumNous#8409
|
No actionable comments were generated in the recent review. WalkthroughGemini 4 Neon is now classified as supporting thinking-level conversion and uses the same effort mappings as Gemini 4 Hydrogen. Tests cover rendering effort values and normalizing a native thinking budget. ChangesGemini 4 Neon thinking-level conversion
Estimated code review effort: 2 (Simple) | ~10 minutes Change: Bug fix · Severity of issue fixed: Medium Suggested reviewers: Merge Risk:
|
Agent: DeepSeek Harness unknown · deepseek-flash · WSL2 Ubuntu · 2026-11-20
Closes #8409
Request summary
The user reported that cross-protocol OpenAI Chat, OpenAI Responses, and Claude requests to
gemini-4-neondropgenerationConfig.thinkingConfig.thinkingLevel. The user asked for the missing mapping to be reported to this repository. The expected result is the same mappinggemini-4-hydrogenhas:gemini-4-neonis level-based,minimalmaps tolow, andlow,medium,high,xhigh,maxpass through.Summary
Add
gemini-4-neonto the Gemini 4 thinking-level mapping that cross-protocol conversion uses. This change is AI-assisted.Evidence
GOWORK=off go test ./relayconvert/reasoning/ -run 'TestRenderGeminiMapsAndClampsInsteadOfRejecting|TestNormalizeGeminiThinkingConfigRewritesNativeConfig' -count=1fails on the new neon cases:ok github.com/QuantumNous/new-api/relaykit/relayconvert/reasoning 0.005s.GOWORK=off go test ./relayconvert/reasoning/ -count=1printsok.GOWORK=off go build ./...exits 0.GOWORK=off go test ./...reports every relaykit package ok.Merge Danger
Door: two-way
Blast Radius: relay
Summary by CodeRabbit