Unverified50% confidenceFactExact time
HumanEval包含164道编程题，MBPP包含约500道编程题
1
Sources
50%
Confidence
Long-term
Relevance
7/6/2026
First Seen
Sources
GPT-5.5 Codex Reasoning-Token Clustering Explained: Causes of Performance Regression and How to Respond
hackernewshackernews7/4/2026
Related Claims
VerifiedMMLU contains approximately 15,000 multiple-choice questions across 57 subject areas; HumanEval, designed by OpenAI, contains 164 Python programming problems; MATH contains 12,500 competition math problems; and GPQA covers graduate-level science questions.71% similarUnverifiedMBPP 收录了 500 道众包编程题，GSM8K 包含 8500 道小学数学应用题（需 2-8 个推理步骤）66% similarUnverifiedThe community ECC solution contains 63 Agents, all in MD file format, of which 6 use Opus for deep reasoning, 55 use Sonnet for everyday tasks, and 1 uses Haiku.58% similarUnverifiedThe course's accompanying documentation exceeds 40,000 characters and is structured around six major hands-on cases with six live-streaming sessions.57% similarUnverifiedKimi K2.6 supports releasing 300 parallel sub-agents at once and can generate more than 100 files from a single prompt.56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/123358API
curl https://kongchang.com/api/v1/knowledge/claims/123358MCP
get_claim(id=123358)