Unverified50% confidenceFactExact time
The performance of current state-of-the-art models on the highest difficulty tier of FrontierMath is typically only in the single-digit to double-digit percentage range.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/12/2026
First Seen
Valid until: 12/11/2026
Sources
Related Entities
Related Claims
UnverifiedThe L1 vector router achieves an accuracy of over 95% on the top-20 high-frequency intents, only 1.5 percentage points below the L2 large model, but with approximately 30× lower latency and cost.67% similarUnverified测试数据表明RNN头与马尔可夫头的草稿通过率差距仅在1-2%以内，而部署成本差距可达数倍66% similarUnverifiedModels with larger parameter counts and more reasoning steps may consume 10–50 times the GPU compute of lightweight models.66% similarUnverifiedThe host's own testing found that minimal mode completes tasks approximately one-third faster than standard mode, with results comparable to standard mode.66% similarUnverified轻量版模型单次推理的算力消耗往往只有旗舰版本的十分之一甚至更低66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/913881API
curl https://kongchang.com/api/v1/knowledge/claims/913881MCP
get_claim(id=913881)