Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Long-Context Generation on Qwen3 Context length (60K)
Loading...
5.89
Throughput Speedup (α)
SpecPV
2.666
3.503
4.34
5.177
Dec 2, 2025
Throughput Speedup (α)
Draft Accept Length (τ)
Updated 4d ago
Evaluation Results
Method
Method
Links
Throughput Speedup (α)
Draft Accept Length (τ)
SpecPV
Model Size=4B, Partial...
2025.12
5.89
3.29
SpecPV
Model Size=14B, Partia...
2025.12
5.83
3.23
SpecPV
Model Size=4B, Partial...
2025.12
5.82
3.14
SpecPV
Model Size=14B, Partia...
2025.12
5.73
2.96
SpecPV
Model Size=14B, Partia...
2025.12
5.63
3.44
SpecPV
Model Size=4B, Partial...
2025.12
5.52
3.11
SpecPV
Model Size=8B, Partial...
2025.12
5.44
3.06
SpecPV
Model Size=8B, Partial...
2025.12
5.35
3.01
SpecPV
Model Size=8B, Partial...
2025.12
5.31
3.11
EAGLE3-YARN
Model Size=14B, YARN s...
2025.12
2.98
3.31
EAGLE3-YARN
Model Size=8B, YARN sc...
2025.12
2.83
3.22
EAGLE3-YARN
Model Size=4B, YARN sc...
2025.12
2.79
3.18
Feedback
Search any
task
Search any
task