List your company

Throughput on QA (Question Answering)

3.66Speedup FactorEagle3+LTD
0.89361.61182.333.0482Feb 20, 2025Apr 23, 2025Jun 25, 2025Aug 26, 2025Oct 28, 2025Dec 29, 2025Mar 2, 2026

Evaluation results

MethodLinks
Eagle3+LTDModel=Vicuna-13B-v1.3, Temperature=0, Decoding Mode=Greedy2026.033.66——5.37—
Eagle3+LTDModel=DeepSeek-R1-Distill-LLaMA 8B, Temperature=0, Decoding Mode=Greedy2026.033.62——5.37—
Eagle3+LTDModel=Llama-3.1-8B-Instruct, Temperature=0, Decoding Mode=Greedy2026.033.41——5.44—
TALONModel=Llama3-8B, Temperature=02026.013.27—5.09——
TALONModel=Vicuna-13B, Temperature=02026.013.26—5.03——
Eagle3+LTDModel=Vicuna-13B-v1.3, Temperature=12026.033.26——4.97—
TALONModel=DSL-8B, Temperature=02026.013.12—4.66——
OPT-TreeModel=Vicuna-13B, Temperature=02026.013.1—5.22——
Eagle3Model=Vicuna-13B-v1.3, Temperature=12026.033.08——4.64—
EAGLE-3Model=Vicuna-13B, Temperature=02026.013.03—5.24——
Eagle3+Spec++Model=Vicuna-13B-v1.3, Temperature=12026.033.02——4.55—
EAGLE-3Model=Llama3-8B, Temperature=02026.013.01—5.36——
Eagle3+GSModel=Vicuna-13B-v1.3, Temperature=12026.033——4.93—
Eagle3+GTModel=Vicuna-13B-v1.3, Temperature=12026.032.99——4.44—
Eagle3+DDDModel=Vicuna-13B-v1.3, Temperature=12026.032.96——4.11—
Eagle3+DiscoModel=Vicuna-13B-v1.3, Temperature=12026.032.88——4.59—
EAGLE-3Model=DSL-8B, Temperature=02026.012.86—5.02——
Eagle3+C2TModel=Vicuna-13B-v1.3, Temperature=12026.032.83——4.65—
Eagle3+SvipModel=Vicuna-13B-v1.3, Temperature=12026.032.61——4.44—
Eagle3+LTDModel=DeepSeek-R1-Distill-LLaMA 8B, Temperature=12026.032.57——4.39—
TALONModel=Qwen3-8B, Temperature=02026.012.46—3.41——
Eagle3+GTModel=Llama-3.1-8B-Instruct, Temperature=12026.032.46——3.24—
Eagle3Model=DeepSeek-R1-Distill-LLaMA 8B, Temperature=12026.032.44——4.06—
Eagle3+LTDModel=Llama-3.1-8B-Instruct, Temperature=12026.032.39——3.26—
Eagle3+DDDModel=Llama-3.1-8B-Instruct, Temperature=12026.032.36——2.95—
Eagle3+Spec++Model=Llama-3.1-8B-Instruct, Temperature=12026.032.35——3.28—
Eagle3+GSModel=DeepSeek-R1-Distill-LLaMA 8B, Temperature=12026.032.33——4.44—
Eagle3+SvipModel=Llama-3.1-8B-Instruct, Temperature=12026.032.31——3.2—
Eagle3+GSModel=Llama-3.1-8B-Instruct, Temperature=12026.032.31——3.4—
Eagle3Model=Llama-3.1-8B-Instruct, Temperature=12026.032.27——3.06—
Eagle3+C2TModel=Llama-3.1-8B-Instruct, Temperature=12026.032.27——3.06—
Eagle3+DiscoModel=Llama-3.1-8B-Instruct, Temperature=12026.032.2——2.88—
EAGLE-3Model=Qwen3-8B, Temperature=02026.012.09—3.46——
Eagle3+LTDModel=Qwen3-32B, Temperature=0, Decoding Mode=Greedy2026.032.06——2.6—
EAGLE-2(+FR 32k)Backbone=Llama-3-8B, Temperature=1, Implementation=Our Implementation, Feature=FR 32k2025.022.02182.59———
TALONModel=Qwen3-32B, Temperature=02026.011.99—2.64——
Eagle3+LTDModel=Qwen3-14B, Temperature=0, Decoding Mode=Greedy2026.031.98——2.62—
HYDRAModel=Vicuna-13B, Temperature=02026.011.95—2.88——
Eagle3+LTDModel=Qwen3 32B, Temperature=12026.031.88——2.41—
Eagle3+LTDModel=Qwen3 14B, Temperature=12026.031.82——2.55—
EAGLE-3Model=Qwen3-32B, Temperature=02026.011.78—2.66——
EAGLE-2Backbone=Llama-3-8B, Temperature=1, Implementation=Our Implementation2025.021.71155.1———
Eagle3+GSModel=Qwen3 14B, Temperature=12026.031.68——2.57—
MEDUSAModel=Vicuna-13B, Temperature=02026.011.62—2.1——
Eagle3+GSModel=Qwen3 32B, Temperature=12026.031.33——2.46—
SDModel=Vicuna-13B, Temperature=02026.011.2—1.97——
VanillaBackbone=Llama-3-8B, Temperature=1, Implementation=Our Implementation2025.02190.58———
Accelerate2026.02—1.16——138.42
DDDTemperature=0, Backbone=Ling-flash-2.0, Batch Size=1, TP (Tensor Parallelism)=4, steps=4, topk=8, draft_tokens=322026.05—187.23———
DDDTemperature=1, Backbone=Ling-flash-2.0, Batch Size=1, TP (Tensor Parallelism)=4, steps=4, topk=8, draft_tokens=322026.05—180.36———
EAGLE-3Temperature=0, Backbone=Ling-flash-2.0, Batch Size=1, TP (Tensor Parallelism)=4, steps=4, topk=8, draft_tokens=322026.05—198.58———
EAGLE-3Temperature=1, Backbone=Ling-flash-2.0, Batch Size=1, TP (Tensor Parallelism)=4, steps=4, topk=8, draft_tokens=322026.05—188.61———
EVICTTemperature=0, Backbone=Ling-flash-2.0, Batch Size=1, TP (Tensor Parallelism)=4, steps=4, topk=8, draft_tokens=322026.05—225.28———
EVICTTemperature=1, Backbone=Ling-flash-2.0, Batch Size=1, TP (Tensor Parallelism)=4, steps=4, topk=8, draft_tokens=322026.05—208.96———
Fate2026.02—7.83——24.23
HybriMoE2026.02—11.83——15.45
llama.cppspeculative decoding=false2026.02—15.09——11.07
llama.cppspeculative decoding=true2026.02—16.49——10.43
LookaheadTemperature=0, Backbone=Ling-flash-2.0, Batch Size=1, TP (Tensor Parallelism)=42026.05—121.47———
LookaheadTemperature=1, Backbone=Ling-flash-2.0, Batch Size=1, TP (Tensor Parallelism)=42026.05—117.18———
Mixtral Offload2026.02—2.33——70.86
MoE-Infinity2026.02—3.44——46.13
MoE-SpAc2026.02—23.56——8.36
SP-MoE2026.02—4.55——34.62
VanillaTemperature=0, Backbone=Ling-flash-2.0, Batch Size=1, TP (Tensor Parallelism)=42026.05—175.32———
VanillaTemperature=1, Backbone=Ling-flash-2.0, Batch Size=1, TP (Tensor Parallelism)=42026.05—170.78———
vLLM2026.02—2.68——57.43