Unverified50% confidenceFactExact time
Qwen3.8 Flash Next has a native context window of 262K tokens, which can be extended to 1 million tokens using YaRN technology
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/6/2026
First Seen
Valid until: 12/5/2026
Sources
Related Entities
Related Claims
UnverifiedThe 4-bit quantized version (Q4K XL) of Qwen3 Next Flash is approximately 111 GB.77% similarVerifiedQwen3-27B natively supports a long context of approximately 260,000 tokens, which can be extended to the 1,000,000-token scale through technologies such as YaRN.75% similarUnverifiedThe approximately 20:1 ratio between Qwen3 Next Flash's 125 billion total parameters and 6 billion activated parameters represents a fairly aggressive sparsity design.70% similarUnverifiedAlibaba's Tongyi Qianwen officially released Qwen3.8-Flash-Next (Qianwen 308 Flash NEX), with total parameters as high as 181.5B.70% similarUnverifiedQwen3.8-Flash-Next's MoE layers use 512 experts with Top-10 routing, a native context window of 262K, and can scale up to 1 million tokens.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/867075API
curl https://kongchang.com/api/v1/knowledge/claims/867075MCP
get_claim(id=867075)