Unverified50% confidenceFactExact time
BenchLM marks the row containing MiMo-V2.6-Pro as Estimated and gives a conditional interval of 63.92 to 83.37, an interval width of 19.45 points; rows marked as Supported in the same table have interval widths of only 8.2 to 10.0 points.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/19/2026
First Seen
Valid until: 11/2/2026
Sources
Related Entities
Related Claims
UnverifiedData from both BenchLM and Artificial Analysis on October 19 show MiMo-V2.6-Pro ranked first among open-weight models, with BenchLM giving 73.7 points and Artificial Analysis giving an intelligence index of 46.3.71% similarUnverifiedSegFormer-B2 is a medium-sized model with approximately 25M parameters, achieving an mIoU of over 46% on the ADE20K urban scene benchmark.63% similarUnverifiedThe model card gives GLM-5.3-Flash's benchmark scores: 84.3 on Terminal-Bench 2.1 and 63.4 on DeepSWE.62% similarUnverifiedThe usable resolution of LLM assessment scores is about 0.5 points, not 0.1 points.61% similarUnverifiedThe previous generation MiMo-V2.5 PRO scored 26 points on the same leaderboard, meaning MiMo-V2.6 PRO improved by 20 points in a single generation.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1069251API
curl https://kongchang.com/api/v1/knowledge/claims/1069251MCP
get_claim(id=1069251)