Which model leads MathArena Apex (Pass@1)?

Q: Which model leads MathArena Apex (Pass@1)?

DeepSeek-V4-Pro tops the board at 90.2

MathArena Apex (Pass@1) Leaderboard — 9 Models

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

The field · 9 measured

23.490.2

9 models measured, most on the official card harness. DeepSeek-V4-Pro tops the board at 90.2.

9 measured·Lifecycle Updated Oct 28, 2026Score basis

#◆Vendor$/M in$/M out

Leader — ignites whiteField — fill cools with scoreIndependently verified only0 of 9Bars show 23.4–90.2 · ◆ solid = first-party or better · outlined = arm's-length

01DeepSeek-V4-Pro+1 altDeepSeek90.2$0.43/M$0.87/M$0.007

02Hy4Tencent74.2$0.83/M$2.5/M$0.022

03GPT-5.6 SolOpenAI65.6$4.0/M$20.0/M$0.183

04DeepSeek-V4.1-FlashDeepSeek65.6$0.14/M$0.42/M$0.004

05Qwen3.7 MaxAlibaba44.5$2.5/M$7.5/M$0.112

06Hy3-previewTencent38.7$0.14/M$0.56/M$0.009

07Seed2.1ByteDance35.4———

08Seed 2.1 Pro Preview+1 altByteDance31.3———

09Gemini 3 ProGoogle23.4$2.0/M$12.0/M$0.299

◆ 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: 9 models scored · 0 independently verified · 1 vendor cross-reference · 8 vendor-reported · 1 with source disagreement. How these tiers are assigned