Qwen3.5 122B A10B (Reasoning)
Open weightsOpen weightsText122B total / 10B active (MoE)Reasoning
Qwen family model by Alibaba Cloud — reasoning.
About the Qwen family
Alibaba's Qwen — the world's most-forked open family: dense and MoE models from 0.6B to trillion-scale Max tiers, plus Coder, VL (vision), Omni (speech/vision/text) and dedicated reasoning variants (QwQ, Thinking). Qwen 3.5–3.7 span open weights and commercial Plus/Max APIs.
All Qwen modelsHow this one is configured
This configuration runs in reasoning mode (thinks step-by-step before answering); mixture-of-experts with ~122B total / ~10B active parameters per token.
Who makes it
Alibaba CloudChinaAlibaba's cloud arm and home of the prolific open Qwen family.More from Qwen
- Qwen Chat 14B
- Qwen Chat 72B
- Qwen1.5 Chat 110B
- Qwen2 Instruct 72B
- Qwen2.5 Coder Instruct 32B
- Qwen2.5 Coder Instruct 7B
- Qwen2.5 Instruct 32B
- Qwen2.5 Instruct 72B
- Qwen2.5 Max
- Qwen2.5 Turbo
- Qwen3 0.6B (Non-reasoning)
- Qwen3 0.6B (Reasoning)
- Qwen3 1.7B (Non-reasoning)
- Qwen3 1.7B (Reasoning)
- Qwen3 14B (Non-reasoning)
- Qwen3 14B (Reasoning)
- Qwen3 235B A22B (Non-reasoning)
- Qwen3 235B A22B (Reasoning)
- Qwen3 235B A22B 2507 (Reasoning)
- Qwen3 235B A22B 2507 Instruct
- Qwen3 30B A3B (Non-reasoning)
- Qwen3 30B A3B (Reasoning)
- Qwen3 30B A3B 2507 (Reasoning)
- Qwen3 30B A3B 2507 Instruct