Share your thoughts, 1 month free Claude Pro on usSee more

Long-context QA on LongBench Overall v2

42.5Accuracy

Qwen3-Coder-30B

Updated 1mo ago

Evaluation Results

Method	Links
Qwen3-Coder-30B 2026.06		42.5
Qwen3-32B 2026.06		36.8
Qwen3-14B 2026.06		34.2
CONTEXTRL 2026.06		33.2
RL baseline 2026.06		31.8
Base model 2026.06		31.6
CONTEXTRL 2026.06		29.6
Base model 2026.06		27.4
RL baseline 2026.06		27