Qwen3-4B
Qwen
BenchClub cloud
- Pass rate
- 78%
- 21 of 27 checks
- Speed
- 247.9 tok/s
- decode, short prompt
- At 8k context
- 193.4 tok/s
- decode
- Cloud runs
- 3
- speeds within 1.8%
- Coding6/967%
- Tool use6/967%
- Long context9/9100%
- Suite
- [email protected]
- Machine
- NVIDIA L40S · 48 GB
- Engine
- ollama 0.40.14
Headline numbers come only from cloud runs. Runs on members’ own machines are listed below and don’t change them.
LMArena
Not ranked on LMArena
LMArena only ranks the models it serves.
BenchClub runs
Run this modelRun byWhereSpeed
- Lucas OliveiraNov 4, 2026RTX 4060 Ti · 16 GB82.1 tok/s
- Haruto SatoNov 2, 2026Apple M3 · 16 GB25.6 tok/s
Ivy ChenOct 23, 2026L40S · 48 GBCloud249.7 tok/s
- Lucas OliveiraOct 22, 2026RTX 4060 Ti · 16 GB85.2 tok/s
- Priya RamanOct 22, 2026Apple M2 Pro · 32 GB51.8 tok/s
- Tom BeckerOct 20, 2026AMD Ryzen 9 7950X · 64 GB19.7 tok/s
Hana KimOct 18, 2026Apple M4 Pro · 24 GB70.0 tok/s
- Priya RamanOct 18, 2026L40S · 48 GBCloud247.9 tok/s
- Priya RamanOct 16, 2026RTX 4070 Ti SUPER · 16 GBNot counted
- Haruto SatoOct 12, 2026Apple M4 Pro · 24 GB71.4 tok/s
Samir HaddadOct 8, 2026L40S · 48 GBCloud245.2 tok/s
- Arjun MehtaOct 2, 2026Apple M5 Pro · 48 GB80.0 tok/s
PassedFailedNot supported