Leaderboard attribution: community run showing as Anthropic submission

[JoeProAI]

Our PinchBench run (94.8%, model field anthropic/claude-opus-4-6) is displaying on the leaderboard as if it were an official Anthropic submission. It's not -- it's a community run by JoeProAI running OpenClaw on a personal setup (RTX 5080, WSL2).

We previously updated the model field to jobeous/anthropic-claude-opus-4-6 to differentiate it, but it still appears grouped under Anthropic.

Request: Either relabel the entry to show the submitter as JoeProAI / Jobeous, or remove it so we can resubmit with correct attribution. The underlying model is Claude Opus 4.6, but the submission is ours, not Anthropic's.

GitHub: @JoeProAI
PinchBench submission key holder: Joe Gaither

[olearycrew]

@JoeProAI what submission ID is this?

[JoeProAI]

Updated submission ID for our 94.8% run: 4eea97e0-1284-4715-8cdf-5f2290ce40ee

View: https://pinchbench.com/submission/4eea97e0-1284-4715-8cdf-5f2290ce40ee

This is the run we'd like relabeled as a JoeProAI / Jobeous community submission rather than grouped under Anthropic.

[JoeProAI]

Correction on my previous comment -- please disregard the new upload. I accidentally created a duplicate submission.

The submissions that need to be relabeled or removed are:

• 20d3604f-371f-4d5c-9e62-eebfc29d30ae (original 93.8% run)
• 4eea97e0-1284-4715-8cdf-5f2290ce40ee (accidentally re-uploaded just now, can be removed)

Both are community runs by JoeProAI, not official Anthropic submissions. Ideally relabel to show submitter as JoeProAI / Jobeous, or remove and we'll resubmit with correct attribution.

[JoeProAI]

Quick follow-up on this. We noticed a few responses in the project lately that read like they were drafted by AI — which honestly is a perfect example of why attribution matters. It's just so easy and accurate to spot now.

On the actual issue: our run completed all 23 PinchBench tasks with verified results. Current unofficial runs that cherry-pick a single task category can land at the top of the board without representing a full result set. Our 94.8% is across the complete suite.

We'd still appreciate being listed as JoeProAI / Jobeous rather than appearing as an official Anthropic submission. Happy to resubmit if that's easier.

[ScuttleBot]

👋 Hi! I'm @olearycrew's OpenClaw bot, helping triage this issue.

Thanks for the detailed info and the submission IDs — that makes this much easier to track down.

What I understand:

• You ran Claude Opus 4.6 via OpenClaw on your own hardware (RTX 5080, WSL2)
• The leaderboard is grouping it under Anthropic based on the model name, but it should be attributed to JoeProAI / Jobeous as a community submission
• Submissions in question:
  • 20d3604f-371f-4d5c-9e62-eebfc29d30ae — original 93.8% run (relabel)
  • 4eea97e0-1284-4715-8cdf-5f2290ce40ee — duplicate (can be removed)

This is a database/admin fix, not something that can be solved with a PR. Tagging @olearycrew to update the attribution directly.

Joe — once this is fixed, would adding a "submitter" field separate from "model" to the submission flow prevent this in the future? Just thinking ahead for other community runs.