List your company

Language Understanding on MMLU (Score and Relative Gain)

90.8MMLU ScoreDeepSeek V3.2
76.55280.25183.9587.649May 21, 2026

Evaluation results

MethodLinks
DeepSeek V3.2SLM=DeepSeek V3.22026.0590.8—
+PACESLM=Qwen3.5-9B2026.0588.98.6
+PACESLM=Ministral-3-14B2026.0585.46.7
+CESLM=Ministral-3-14B2026.0583.64.5
Gödel AgentSLM=Qwen3.5-9B2026.0583.52.1
+CESLM=Qwen3.5-9B2026.0583.52.1
MIPROv2SLM=Ministral-3-14B2026.0583.34.1
GEPASLM=Qwen3.5-9B2026.0583.21.7
+PESLM=Qwen3.5-9B2026.0583.21.7
+PESLM=Ministral-3-14B2026.0583.24.4
GEPASLM=Ministral-3-14B2026.0582.53.1
+PACESLM=Qwen3-4B2026.0581.86.1
VanillaSLM=Qwen3.5-9B2026.0581.8—
ACESLM=Ministral-3-14B2026.0581.82.2
ACESLM=Qwen3.5-9B2026.0580.5-1.5
ACESLM=Qwen3-4B2026.05803.7
VanillaSLM=Ministral-3-14B2026.0580—
Gödel AgentSLM=Qwen3-4B2026.0579.73.3
+CESLM=Qwen3-4B2026.0579.73.3
MIPROv2SLM=Qwen3.5-9B2026.0579.7-2.5
Gödel AgentSLM=Ministral-3-14B2026.0579.70
+PESLM=Qwen3-4B2026.0578.82.2
VanillaSLM=Qwen3-4B2026.0577.1—
MIPROv2SLM=Qwen3-4B2026.0577.10
GEPASLM=Qwen3-4B2026.0577.10