collabosm rents a GPU in your own Google Colab account to serve Qwen3.8-27B or Flash-Next locally

A Reddit user shared that Google's Gemini Pro subscription includes 200 free Google Colab compute credits monthly, unlocking the Colab G4 GPU (Nvidia RTX Pro 6000 Blackwell Server Edition, 96GB VRAM) for roughly 30+ hours per month. They use it to run Qwen 27B locally with the Cline extension in VS Code at ~50 tokens/second, as a backup when their Antigravity quota runs out.

Social·Community

SWE builds 3x RTX 3090, 64GB VRAM local AI coding PC running Qwen3.8-27B via llama.cpp

A software engineer detailed a DIY workstation with three RTX 3090s (~62GB usable VRAM) built to replace paid OpenAI/Anthropic subscriptions for coding agents. The build runs a Qwen3.8-27B Q8_0 GGUF quant through a self-compiled llama.cpp server with 262k context, at 44-46 t/s dropping to 28-30 t/s as context fills. It shows local multi-GPU setups are now practical for real software engineering workflows, though with nontrivial hardware constraints around PSUs, risers, and cooling.

A software engineer detailed a DIY workstation with three RTX 3090s (~62GB usable VRAM) built to replace paid OpenAI/Anthropic subscriptions for coding agents. The build runs a Qwen3.8-27B Q8_0 GGUF quant through a self-compiled llama.cpp server with 262k context, at 44-46 t/s dropping to 28-30 t/s as context fills. It shows local multi-GPU setups are now practical for real software engineering workflows, though with nontrivial hardware constraints around PSUs, risers, and cooling.

Releases·Model Release

UkisAI Swift-Qwen3.8-27B: -58.3% thinking tokens, 1.95x speed, xhigh accuracy kept

UkisAI post-trained Qwen3.8 27B into Swift-Qwen3.8-27B by using RL to penalize reasoning-marker tokens linked to overthinking, then fixing accuracy with on-policy distillation. The open-source model cuts thinking tokens 58.3% and speeds inference 1.95x while keeping xhigh accuracy, and community benchmarks independently confirm the gains. A thank-you post reports 100k+ downloads, #1 finetune and #9 model on HuggingFace Trending, with a Swift1.5 checkpoint and Swift Qwen3.8 Flash Next upcoming.

UkisAI post-trained Qwen3.8 27B into Swift-Qwen3.8-27B by using RL to penalize reasoning-marker tokens linked to overthinking, then fixing accuracy with on-policy distillation. The open-source model cuts thinking tokens 58.3% and speeds inference 1.95x while keeping xhigh accuracy, and community benchmarks independently confirm the gains. A thank-you post reports 100k+ downloads, #1 finetune and #9 model on HuggingFace Trending, with a Swift1.5 checkpoint and Swift Qwen3.8 Flash Next upcoming.

Releases·Tool Release

mlx-dspark v0.10.0 speeds Qwen3.8-27B up to 3× on Apple Silicon

mlx-dspark, an MLX port of DeepSeek's DSpark speculative-decoding drafters, now supports Qwen3.8-27B via RadixArk's drafter, achieving up to ~3× speedups on Apple Silicon. Benchmarks on an M4 Pro show 2.45× mean speedup for 8-bit targets, with code runs hitting 3.18×.

mlx-dspark, an MLX port of DeepSeek's DSpark speculative-decoding drafters, now supports Qwen3.8-27B via RadixArk's drafter, achieving up to ~3× speedups on Apple Silicon. Benchmarks on an M4 Pro show 2.45× mean speedup for 8-bit targets, with code runs hitting 3.18×.

News·Incident

Qwen3.8-Flash-Next generates signed Alibaba Cloud OSS URLs during Amazon browsing, spurring exfiltration concerns

A Reddit user running Qwen3.8-Flash-Next on a Mac Studio observed the model issue a browser_navigate call to a signed URL on aliyuncs.com, Alibaba's cloud storage domain, during unrelated Amazon research. They cite a similar Hacker News report from a Qwen3.8-27B user 45 days earlier. Possible explanations range from benign hallucination of Alibaba training traces to data exfiltration; no root cause is confirmed.

A Reddit user running Qwen3.8-Flash-Next on a Mac Studio observed the model issue a browser_navigate call to a signed URL on aliyuncs.com, Alibaba's cloud storage domain, during unrelated Amazon research. They cite a similar Hacker News report from a Qwen3.8-27B user 45 days earlier. Possible explanations range from benign hallucination of Alibaba training traces to data exfiltration; no root cause is confirmed.