Unverified50% confidenceFactExact time
Meta admits that its internally run competitor tests "use common settings where feasible, but are still best-effort: prompts, tools, and runtime may not be optimized for proprietary models".
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedMeta选择在Product Hunt而非自有渠道首发Pocket，暗示其仍处于实验性阶段77% similarUnverifiedThe Meta model card footnote explains: the score selection method prioritizes comparable third-party results, otherwise uses the highest of internal validation runs, self-reported scores, or official leaderboard scores; competitor runs are best-effort and not optimized for proprietary models; the SWE-Marathon comparison involves different agent harnesses; Wonder Full's 57.5 on SWE-Marathon is reported with xhigh reasoning effort.75% similarUnverifiedUnder a MoE architecture, more parameters do not necessarily yield better results, and actual testing on the target task remains the most reliable basis for model selection.72% similarUnverifiedMeta imposes test mode restrictions on new applications: before submitting for review, messages can only be sent to and received from pre-added test numbers within a whitelist.70% similarUnverifiedThe study validates across multiple benchmarks that, under controlled candidate pool and team size conditions, the framework consistently outperforms quality-only baselines.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1011215API
curl https://kongchang.com/api/v1/knowledge/claims/1011215MCP
get_claim(id=1011215)