Which model leads CMath (EM)?

Q: Which model leads CMath (EM)?

DeepSeek-V4-Flash-Base tops the board at 93.6

CMath (EM) Leaderboard — 7 Models Ranked

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

The field · 7 measured

77.393.6

7 models measured, most on the official card harness. DeepSeek-V4-Flash-Base tops the board at 93.6.

7 measured·Lifecycle Updated Oct 8, 2026Score basis

#◆Vendor$/M in$/M out

Leader — ignites whiteField — fill cools with scoreIndependently verified only0 of 7Bars show 77.3–93.6 · ◆ solid = first-party or better · outlined = arm's-length

01DeepSeek-V4-Flash-Base+2 altsDeepSeek93.6———

02DeepSeek-V3.2-Base+3 altsDeepSeek92.6———

03DeepSeek-V4-Pro-Base+3 altsDeepSeek90.9———

04DeepSeek-V3DeepSeek90.7$0.24/M$0.90/M$0.006

05Qwen2.5 Instruct 72BAlibaba84.5$0.47/M$0.49/M$0.006

06DeepSeek-V2DeepSeek78.7———

07Llama 3.1 Instruct 405BMeta77.3———

◆ 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: 7 models scored · 0 independently verified · 2 vendor cross-reference · 5 vendor-reported · 0 with source disagreement. How these tiers are assigned