Which model leads CRUXEval-I (input prediction)?

Q: Which model leads CRUXEval-I (input prediction)?

Hy3-preview-Base tops the board at 71.19

CRUXEval-I (input prediction) Leaderboard — Model Scores

Pro Mode — raw scores, one benchmark under the lensofficial card harness

The field · 11 measured

52.571.19

11 models measured, most on the official card harness. Hy3-preview-Base tops the board at 71.19.

11 measured·Lifecycle Updated Oct 8, 2026Score basis

#◆Vendor$/M in$/M out

Leader — ignites whiteField — fill cools with scoreIndependently verified only0 of 11Bars show 52.5–71.19 · ◆ solid = first-party or better · outlined = arm's-length

01Hy3-preview-Base+1 altTencent71.19———

02GLM-4.5-Base+1 altZ.ai68.51———

03Kimi K2 Base+2 altsMoonshot68.01———

04DeepSeek-V3-Base+1 altDeepSeek67.79———

05MiMo V2 Flash BaseXiaomi67.53P———

06DeepSeek-V3DeepSeek67.3$0.24/M$0.90/M$0.008

07DeepSeek-V3.2-Exp-BaseDeepSeek63.93P———

08DeepSeek-V3.1-BaseDeepSeek62.13P———

09Qwen2.5 Instruct 72BAlibaba59.1$0.47/M$0.49/M$0.008

10Llama 3.1 Instruct 405BMeta58.5———

11DeepSeek-V2DeepSeek52.5———

◆ solid = independently verified · aggregator attested · vendor attributed — outlined = cross-reference · source-attributed. Dimmed ◆ = confidence inferred. “+N alts” = alternate disclosures in the drawer with why-demoted. Prices per million tokens; $/point = input price ÷ score. Every row opens its model dossier.

Verification: 11 models scored · 0 independently verified · 7 vendor cross-reference · 4 vendor-reported · 0 with source disagreement. How these tiers are assigned