wowsuch ai
Live

UkisAIModel

Swift 1.5 Qwen3.8-Flash-Next

Swift 1.5 Qwen3.8-Flash-Next is UkisAI’s reasoning-efficient derivative of Qwen3.8-Flash-Next, released as a BF16 checkpoint with quantized variants. UkisAI reports 63.4% fewer median thinking tokens and 1.8× speedup at xhigh, with less than 1% accuracy loss compared with the base model.

What it does

Use it for reasoning, coding, and long-horizon agent tasks with reduced reasoning-token use, either from downloadable weights or through UkisAI’s API.

Availability
BF16 checkpoint and quantized variants available for download; OpenAI-compatible API at ukisai.com/api/flash-next/v1
Pricing model
Free for research purposes via API; enterprise licensing referenced
Deployment
Hosted API or downloadable checkpoint for local-compatible runtimes
Privacy
Not established
Licence
Enterprise licensing referenced; licence terms not specified
Capability
analysiscode generationagents
Modality
textcode
Access
APIopen weights
Deployment
hostedon-premises
Pricing
freeenterpriseunknown
Target user
developerresearcher

Evidence

5 claims
  • The model uses 63.4% fewer median thinking tokens and is 1.8x faster than the base on xhigh, with accuracy loss under 1%. [1]
    It uses **63.4% fewer thinking tokens**, with a **1.8x speed up** while keeping the **accuracy loss <1%** vs base on xhigh.
    Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • The release is post-trained for coding and long-horizon agent work, including personal agents, terminal use, and software engineering. [1]
    work such as personal agents, terminal use and software engineering.
    Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • The BF16 checkpoint is about 360 GB and requires a multi-GPU node. [1]
    The BF16 checkpoint is about 360 GB, so it needs a multi-GPU node.
    Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • Quantized versions are available in AWQ, AutoRound, NVFP4, GGUF, and GSQ-RCO GGUF formats. [1]
  • The OpenAI-compatible API is free for research purposes and requires no API key. [1]
    It is **free for research purposes** and needs no
    Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.

Sources

1
1
Hugging FaceOwner-controlled
https://huggingface.co/ukisai/Swift-Qwen3.8-Flash-Next Retrieved 2026-10-28 01:03 UTC
Verified

Versions

3
VersionEffectiveChangeChanged fieldsState
32026-10-23 05:27 UTCevidence updatescoreCurrent
22026-09-26 06:49 UTCmaterial changemodality, access, deployment, pricing, availability, pricing model, deploymentCurrent
12026-09-25 07:48 UTCpublicationNoneCurrent

In briefs

1

Open in the catalog

Original catalog content is licensed under CC BY 4.0, attributed to wowsuchai. Source quotations and linked content keep their original rights.