Polymer Science Response Quality Evaluation on External Polymer Benchmarks Blk, ChemD, Llml 1.0 (test)
4.93Blk ScorePolyLM (Phi-4-14B)
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| PolyLM (Phi-4-14B)Category=PolyLM, Backbone=Phi-4-14B, Chain-of-Thought=false2026.01 | 4.93 | 6.49 | 3.34 | 4.92 | 79.56 | |
| GPT-5Category=Closed-source, Chain-of-Thought=false2026.01 | 4.8 | 6.75 | 3.9 | 5.15 | 94 | |
| SonnetCategory=Closed-source, Chain-of-Thought=false2026.01 | 4.52 | 6.63 | 3.61 | 4.92 | 80.08 | |
| PolyLM (Phi-4-14B)Category=PolyLM, Backbone=Phi-4-14B, Chain-of-Thought=true2026.01 | 4.43 | 6.33 | 3.52 | 4.76 | 85.01 | |
| PolyLM (Qwen-2.5-14B)Category=PolyLM, Backbone=Qwen-2.5-14B, Chain-of-Thought=true2026.01 | 4.14 | 6.28 | 3.74 | 4.72 | 91.15 | |
| PolyLM (Qwen-2.5-14B)Category=PolyLM, Backbone=Qwen-2.5-14B, Chain-of-Thought=false2026.01 | 4.09 | 6.41 | 3.43 | 4.64 | 84.77 | |
| Qwen-2.5-14BCategory=Off-the-shelf, Chain-of-Thought=false2026.01 | 4.02 | 6.08 | 3.56 | 4.55 | 79.92 | |
| GPT-4oCategory=Closed-source, Chain-of-Thought=false2026.01 | 3.76 | 6.37 | 2.71 | 4.28 | 73.92 | |
| 4o-MiniCategory=Closed-source, Chain-of-Thought=false2026.01 | 3.58 | 6.41 | 2.66 | 4.21 | 69.17 | |
| ChemLLM-7BCategory=Domain aligned, Chain-of-Thought=false2026.01 | 2.99 | 5.76 | 2.62 | 3.79 | 56.08 | |
| Phi-4-14BCategory=Off-the-shelf, Chain-of-Thought=false2026.01 | 2.86 | 5.6 | 2.96 | 3.81 | 56.17 | |
| ether0-24BCategory=Domain aligned, Chain-of-Thought=false2026.01 | 1.63 | 1.37 | 1.14 | 1.38 | 14.83 | |
| LlaSMol-8BCategory=Domain aligned, Chain-of-Thought=false2026.01 | 1.54 | 4.75 | 1.16 | 2.48 | 29.25 |