Architecture Highlights

• 80B total parameters, 3B activated (A3B)
• 48 layers with hybrid layout
• 512 experts with only 10 activated per token
• Context length: 262,144 tokens
• No thinking mode support (instruct-only)

Performance

Delivers performance comparable to much larger models while maintaining exceptional efficiency:

• Outperforms Qwen3-32B with 10x inference throughput for long contexts
• Matches Qwen3-235B-A22B on many benchmarks with significantly lower computational requirements
• Superior ultra-long-context handling up to 256K+ tokens

qwen3-next-80b

Forked from qwen/qwen3-next-80b