Which model leads PIQA (Acc.)?

Q: Which model leads PIQA (Acc.)?

Llama 3.1 Instruct 405B tops the board at 85.9

PIQA (Acc.) Leaderboard — Model Scores

Pro Mode — raw scores, one benchmark under the lensthird-party eval harness

The field · 9 measured

82.2185.9

9 models measured, most on the third-party eval harness. Llama 3.1 Instruct 405B tops the board at 85.9.

9 measured·Lifecycle Updated Oct 9, 2026Score basis

#◆Vendor$/M in$/M out

Leader — ignites whiteField — fill cools with scoreIndependently verified only0 of 9Bars show 82.21–85.9 · ◆ solid = first-party or better · outlined = arm's-length

01Llama 3.1 Instruct 405BMeta85.9VENDOR———

02DeepSeek-V3DeepSeek84.7VENDOR$0.24/M$0.90/M$0.007

03DeepSeek-V2DeepSeek83.9VENDOR———

04Nemotron 3 Super 120B A12BNVIDIA83.9$0.30/M$0.90/M$0.007

05Nemotron 3.5 LightningNVIDIA83.35$0.06/M$0.20/M$0.002

06NVIDIA Nemotron 3 Nano 30B A3BNVIDIA82.64$0.05/M$0.20/M$0.002

07Qwen2.5 Instruct 72BAlibaba82.6VENDOR$0.47/M$0.49/M$0.006

08Qwen3.5 35B A3BAlibaba82.32$0.25/M$2.0/M$0.014

09Gemma 4 26B A4BGoogle82.21$0.07/M$0.34/M$0.002

◆ 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 · 4 vendor cross-reference · 5 vendor-reported · 0 with source disagreement. How these tiers are assigned