Qwen3 Coder Next (non-reasoning) tokens per second
Live benchmark data for Qwen3 Coder Next (non-reasoning) on Ollama Cloud. Speed, latency, and reliability — benchmarked every ~10 minutes.
- Latest TPS
- 103.2 tok/s
- 24h avg TPS
- 89.7 tok/s
- TTFT
- 335ms
- 24h reliability
- 100%
- Last tested
- 9m ago
Tokens per second over time
Time to first token over time
Reliability by day & hour (7d)
Each cell shows the percentage of successful benchmark runs in that hour. Green = all ok, red = mostly failing.
Speed is output tokens ÷ server-measured generation time. Each benchmark fires one streaming chat completion with a fixed ~300-token prompt, every ~10 minutes. Full methodology →