[KongchangAI]
Unverified50% confidenceBenchmarkExact time

Running OpenVLHarness with GPT-6 Luna at medium reasoning effort, its performance exceeds that of the base model, Codex, and Visual Sketchpad at the highest reasoning effort, leading by 7.7–8.5 points.

1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/20/2026
First Seen
Valid until: 1/18/2027

Sources

OpenVLHarness: How Multimodal Memory Reshapes Tool Use in Visual Agents

Related Entities

GPT-6 LunamodelOpenVLHarnessproductCodexproductVisual Sketchpadproduct

Related Claims

VerifiedGPT-6 Sol and GPT-6 Luna deliver strong performance at lower cost and have pushed the Pareto frontier on the OfficeQA Pro benchmark74% similarVerifiedGPT-6.1 Sol performs best on the DeepSWE benchmark at the high reasoning effort level73% similar

Cite This Claim

Stable URI
https://kongchang.com/claim/1073235
API
curl https://kongchang.com/api/v1/knowledge/claims/1073235
MCP
get_claim(id=1073235)