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

Home/Benchmarks

Conversation on ShareGPT

2.01Throughput

HiSpec

Updated 7d ago

Evaluation Results

Method	Links
HiSpec 2025.10		2.01	100
SPRINTER 2025.10		1.2	40

SOTA Paper

HiSpec

HiSpec: Hierarchical Speculative Decoding for LLMs

Dataset

ShareGPT

Follow for update

@wizwand_team Discord

Related Benchmarks

Text Generation on ShareGPT Summarization on CNN/DM (Speedup vs AR)Code Generation on HumanEval (Speedup vs. AR)Mathematical Reasoning on GSM8K (Speedup vs. AR)Summarization on XSum (Speedup vs. AR)

© 2026 wizwand

Blog Contact Changelog Swarm

Privacy Policy Terms of Service FAQs Swarm Docs