Hi everybody, we post-trained Qwen 3.8 27B to be more efficient by figuring out which tokens were linked to overthinking and penalizing them without "attacking" the reasoning length directly then fixed the accuracy with a bit of secret sauce (hint On-Policy Distillation) and achieved great results (-58% length, 1.95x speed up, <1% accuracy loss) so we wanted to open-source it and hear the feedback of the community.

This is the link to the model: https://huggingface.co/ukisai/Swift-Qwen3.8-27b / It's at 80k downloads in 3 days with independent evals here: https://www.reddit.com/r/LocalLLaMA/comments/1wg7dd5/ukisai_...

We are also providing a Free Research Purpose API (OpenAI compatible), courtesy of Nvidia who were kind enough to provide us with the GPUs. It's limited at 5RPM. https://ukisai.com/api/swift/v1/models

We also made a GGUF (Q1-Q8): https://huggingface.co/ukisai/Swift-Qwen3.8-27B-GGUF and there's also a few nice community quants with even lower/higher precision (Bartowski: https://huggingface.co/bartowski/ukisai_Swift-Qwen3.8-27b-GG...). The community also created amazing MLX, NVFP4, W4A16 and Uncensored versions you can find on Huggingface.

IMPORTANT: Our training approach is not a replacement for the reasoning effort settings, chat templates or token caps but is complementary and targets a completely separate issue (overthinking and "anxiety-like" reasoning loops prior seen in PTQ, but as far as we identified also prominent in BF16 of this size class LLMs as well). Contrary to popular belief, these specific patterns do not contribute to answer quality when properly targeted. (our thesis being: reasoning length IS extremely important and should NOT be shortened by force, but rather optimized). This is also demonstrated bellow in our xhigh vs medium effort benchmark. The goal is to keep xhigh accuracy while reducing only the unnecessary part of thinking.

The TLDR of our thought process, research, training and a link to the Meta paper that inspired us is in the first comment.

The benchmarks:

Swift-27B vs Qwen3.8-27B (BF16, all benchmarks ran x5, thinking effort xhigh)

GPQA-Diamond: 88.4% -> 88.3%, 58% fewer median tokens

LiveCodeBench v6: 76.8% -> 81.6% (+4.8pp, due to default truncation in LCB it is not performance gain), 46% fewer median thinking tokens

Terminal-Bench 2.1: 66.7% -> 65.8%, 39% fewer median tokens

MMLU-Pro: 85.5% -> 85.0%, 28% fewer median tokens

C-Eval: 90.0% -> 90.6%, 19% fewer median tokens

IFBench: 73.5% -> 71.8%, 51% fewer median tokens

AIME 2026: 98.7% -> 94.0%, 50% fewer median tokens

HMMT (Nov 2025): 99.3% -> 96.0%, 46% fewer median tokens

ERQA (vision): 67.5% -> 66.3%, 55% fewer median tokens

Token savings hold at every reasoning effort (mean thinking reduction): xhigh 41%, medium 23%, low 26% (albeit with accuracy loses of 1-4% on medium and 1-2% on low which we need further testing for)

Swift at xhigh vs the base's own effort settings on GPQA-Diamond (198 questions x 5 seeds):

Base xhigh: 88.4%, 6,642 median tokens

Swift xhigh: 88.3%, 2,771 median tokens

Base medium: 84.1%, 1,753 median tokens

So Swift keeps xhigh accuracy at under half the tokens, and beats base-medium by 4pp at roughly 1.6x its tokens.

End note:

While we are keen on complete open-source, we still need to keep a part of our training and data private, being a new lab. The license is not Apache 2.0, but it only affects companies >$1M. We hope this does not pose a problem for the community and are open to feedback on it.

We want to contribute as much as possible to the community and would really appreciate feedback on our work, quantization or Swift model requests. We are working on Swift 3.8 Flash Next and have so far gotten up to -53% thinking token usage. We have strong indicators our methodology is reproducible on other model families as well and are asking the community which ones you want us to optimize next.

Show HN: OmnisBench, a re-gradable, open LLM routing benchmark on fresh tasks

Fourteen studies and every walk-forward fold validated a hindsight stop

Better Icon and Label Alignment

Show HN: Aplexica – resume a Claude Code conversation in Codex, and back

BurpSqueezer – Turn Burp Suite XML Dumps into Compact LLM-Ready Markdown

WordPress Studio Is Now Agentic

Voila: Ship Web Apps Straight from Your LLM Agent

Show HN: Swift-Qwen3.8-27B, -58.3% thinking, x1.95 speed, accuracy of xhigh

Why Mechanical Sympathy? (2011)

CobbleDB: Perplexity developed storage engine using 2 engineers

Survival Guide for a Censored Internet

The Narrow Corridor — The Permanent Struggle for Liberty (2019)

FoxSchema: Database schema diff and migration tool

We know what a world without work looks like

Dream-RSI: Recursive Self-Improvement Through Evolving Worlds

TileDB Rebrands as Tile.ai

Apple Copland D11E4 booting in your Browser

Combinators and the Story of Computation (2025)

Show HN: Zoom Meeting for Agent Sessions

The state of AI in 2026 [pdf]

The mascots, design and visual history of Wikipedia

Heather Cox Richardson

How and Why We Bought 4x DGX Sparks

Show HN: Ctxwitch – Git tells you what changed; this tells you what it'll do

The Frontier Act Is Congress' Best AI Regulation Bill Yet

Switched to Claude Code's open-source rival, stopped worrying about rate limits

Show HN: Ori, an agentic runtime for infrastructure that must ask before it acts

BBC Operations Dealt with 9/11 [correct URL]

Mojo 1.0 and the Open Road Ahead [video]

Be the first human to review your own code