Unverified50% confidenceBenchmarkExact time
Experiments on Qwen3 models for mathematical reasoning tasks show that MTA-OPSD consistently outperforms existing OPSD methods, with an average gain of up to +5.9 relative to the base model.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/30/2026
First Seen
Valid until: 1/28/2027
Sources
Related Entities
Related Claims
UnverifiedRelative to the open-source base model Qwen3.5-4B, OPRD improves the average task resolution rate by 11.3 percentage points across the three benchmarks SWE-bench Verified, SWE-bench Pro, and SWE-bench Multilingual.66% similarUnverifiedOn agentic coding benchmarks such as SWE Bench Pro, Qwen 3.8 reportedly surpasses Opus 4.6 Max on some tasks.65% similarUnverifiedQAOA maps combinatorial optimization problems to Ising Hamiltonians, whose ground states correspond to optimal solutions.63% similarUnverifiedThe Top3 configuration, which uses the fewest metrics, achieved the highest average performance, indicating that fewer metrics actually yield higher performance under this experimental setting.62% similarUnverifiedIn complex reasoning tasks, the number of reasoning tokens can far exceed the number of final output tokens, which is why the usage cost of the o1/o3 series is typically higher than that of GPT models at the same tier.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1115272API
curl https://kongchang.com/api/v1/knowledge/claims/1115272MCP
get_claim(id=1115272)