Model techniques map
Techniquesinference & servingdecoding strategy

specific method · filed under inference & serving

NEXTN speculative decoding

A named speculative-decoding mode associated in the evidence with a multi-step-trained MTP head and draft tokens.

Also called native NEXTN speculative decoding, NEXTN speculative decoding algorithm, NEXTN.

sources
4
models
3
labs adopt it
2
strongest
used

How sources treat it

One count per evidence span, weakest treatment to strongest.

optional 2used 2

Documented in

Evidence

4 spans quoted from the sources, strongest treatment first.

--speculative-algorithm NEXTN --speculative-num-steps 3 --speculative-eagle-topk 1 --speculative-num-draft-tokens 4

usedinference servingin Hy4-previewTencent Hunyuan

the model ships a 1-layer MTP head trained with multiple steps, exposed by SGLang as native NEXTN speculative decoding.

usedinference servingin Qwen3.8-Flash-NextQwen

--speculative-algo NEXTN --speculative-num-steps 3 --speculative-eagle-topk 1 --speculative-num-draft-tokens 4

optionalinference servingin Qwen3.5-122B-A10BQwen

--speculative-algo NEXTN --speculative-num-steps 3 --speculative-eagle-topk 1 --speculative-num-draft-tokens 4

optionalinference servingin Qwen3.5-35B-A3BQwen

Filed alongside

Other methods under inference & serving :: decoding strategy.