Deterministic Stack Engine

Best AI Stack for open-weight reasoning

Highest reasoning signal among models you can self-host or buy as open weights.

Objective Weight Distribution

Elo:30%LiveBench:30%GPQA:25%Out $:15%

Recommended PickScore 73.9 / 100

DeepSeek V4 Pro

DeepSeek · Open weights

Elo: 1,536 (Top 22%)LiveBench: 70.1% (Top 17%)GPQA: 84% (Top 18%)Out $: $3.96/1M (Top 66%)

Output: $3.9600000000000004/1MSWE-bench: 71.6%

View full model fact sheet

Full Stack Leaderboard

Ranked alternatives optimized for open-weight reasoning based on objective benchmark weighting.

#2DeepSeek V4 Flash(DeepSeek)

Score 70.4

Elo1,490LiveBench67.8%GPQA81.6%Out $$0.28/1M

Rank #2 in stackvs #1

#3Qwen QwQ 32B(Alibaba)

Score 70.0

Elo1,495LiveBench68%GPQA83.2%Out $$1.2/1M

Rank #3 in stackvs #1

#4GLM-5.2(Zhipu)

Score 58.0

Elo1,492LiveBench65.2%GPQA79.5%Out $$3.74/1M

Rank #4 in stackvs #1

#5MiniMax M2.5(MiniMax)

Score 53.5

Elo1,478LiveBench64.4%GPQA76.2%Out $$0.9/1M

Rank #5 in stackvs #1

#6Llama 4 Maverick(Meta)

Score 46.9

Elo1,468LiveBench62.8%GPQA74.6%Out $$0.8/1M

Rank #6 in stackvs #1

#7Qwen3 235B(Alibaba)

Score 45.5

Elo1,440LiveBench63.5%GPQA76.8%Out $$1.82/1M

Rank #7 in stackvs #1

#8Qwen 2.5 Coder 32B(Alibaba)

Score 45.1

Elo1,425LiveBench64.8%GPQA74%Out $$0.72/1M

Rank #8 in stackvs #1

#9DeepSeek R1(DeepSeek)

Score 35.8

Elo1,358LiveBench59%GPQA79.8%Out $$2.5/1M

Rank #9 in stackvs #1

#10Llama 4 Scout(Meta)

Score 34.7

Elo1,412LiveBench59.6%GPQA71.2%Out $$0.3/1M

Rank #10 in stackvs #1