Life & fun · alternatives · 2026
24 alternatives to Running 104GB Qwen3.8-Flash-Next on 48GB Mac with at ~12 tok/s
I built slotstream, a way to run Qwen3.8-Flash-Next 4-bit on a low-memory mac starting from 16GB, a 125B parameter model that would need 100GB+ memory/RAM, thanks to expert-offloading/ssd-streaming. Easy to install/update, and mac-native using MLX and Swift. It ships with auto-mode, which makes a good tradeoff between memory usage and speed. I'll be implementing and porting the MTP module for speculative decoding next Local models really are the future of computing!
SlotStream enables users to run the Qwen3.8-Flash-Next 125B mixture-of-experts model on Mac computers with limited RAM by streaming model components from storage to memory as needed. Built with MLX and Swift, it uses… Below are 24 products that do a similar job, ranked by how close each is in meaning and then by launch-day votes. Prices are shown for the 1 we have checked on their own sites, 1 of them free or with a free tier.
- 1
Contribute to Mega4alik/ollm development by creating an account on GitHub.
2025 · github.com · its alternatives →
- 2
Swiftlet is a Swift and Metal runtime that runs large Qwen Mixture-of-Experts models locally on Apple devices by streaming expert weights from storage, enabling 35B and 80B models to run with low RAM, including on iPhone. - leonickson1/Swiftlet
Aug 2026 · github.com · its alternatives →
- 3
Qwen3.8-Flash-Nextfree tier▲108The open-weight preview of Qwen4
Aug 2026 · qwen.ai · its alternatives →
- 4
Hi HN, I built a specialized inference engine for running 4-bit Gemma 4 26B-A4B-IT on any M-series Mac using about 2 GB of RAM. It is called TurboFieldfare and is written in Swift and Metal. I have always adored on-device AI. It feels like magic that you can run a powerful NN on your Mac or iPhone. So I wanted to push the limits a bit and run a model whose weights don’t fit in memory. The model’s 4-bit quantized weights occupy roughly 14 GB, which makes running it with conventional inference tools almost impossible on an 8 GB or even 16 GB Mac once the OS, applications, and KV cache are…
Jul 2026 · github.com · its alternatives →
- 5
Hi HN! I'm just sharing a project I've been working on during the LLM Efficiency Challenge - you can now finetune Llama with QLoRA 5x faster than Huggingface's original implementation on your own local GPU. Some highlights: 1. Manual autograd engine - hand derived backprop steps. 2. QLoRA / LoRA 80% faster, 50% less memory. 3. All kernels written in OpenAI's Triton language. 4. 0% loss in accuracy - no approximation methods - all exact. 5. No change of hardware necessary. Supports NVIDIA GPUs since 2018+. CUDA 7.5+. 6. Flash Attention support via Xformers. 7. Supports 4bit and 16bit…
2023 · github.com · its alternatives →
- 6
Qwen3.8-Max▲277Qwen3.8-Max + MCP for coding on your local machine, without paying for Qwen Code. Qwen3.8-Max itself runs in the cloud through Qwen Studio — this setup just gives it access to your local files and terminal through MCP. My take: it is slower than Codex and Claude Code, especially in thinking mode, but it does work well. MCP adds some latency too since there’s an extra layer between the model and your machine. In fast mode, it gets noticeably closer. I wouldn’t put it in the same category as Fable or Opus-5. Personally, I’d place it closer to the next Opus 4.7 level in terms of coding ability.…
Aug 2026 · qwen.ai · its alternatives →
- 7
Qwen3.5 Small▲3550.8B-9B native multimodal w/ more intelligence, less compute
Mar 2026 · huggingface.co · its alternatives →
- 8
Ollama v0.19▲410Massive local model speedup on Apple Silicon with MLX
Apr 2026 · ollama.com · its alternatives →
- 9
Fine-tune LLMs from one YAML. Layer streaming trains an 8B model on a 4 GB laptop GPU. - MakazhanAlpamys/Soup
Aug 2026 · github.com · its alternatives →
- 10
Metal-first MoE inference for Apple Silicon with bounded SSD expert streaming and local OpenAI Chat, Responses, and Anthropic Messages endpoints. - hebrus-labs/hebrus
Jul 2026 · github.com · its alternatives →
- 11
About six months ago, I started working on a project to fine-tune Whisper locally on my M2 Ultra Mac Studio with a limited compute budget. I got into it. The problem I had at the time was I had 15,000 hours of audio data in Google Cloud Storage, and there was no way I could fit all the audio onto my local machine, so I built a system to stream data from my GCS to my machine during training. Gemma 3n came out, so I added that. Kinda went nuts, tbh. Then I put it on the shelf. When Gemma 4 came out a few days ago, I dusted it off, cleaned it up, broke out the Gemma part from the Whisper…
Apr 2026 · github.com · its alternatives →
- 12
- 13TF
I’d originally launched my app: Private LLM[1][2] on HN around 10 months ago, with a single RedPajama Chat 3B model. The app has come a long way since then. About a month ago, I added support for 4-bit OmniQuant quantized Mixtral 8x7B Instruct model, and it seems to outperform Q4 models at inference speed and Q8 models at text generation quality, while consuming only about 24GB of RAM[3] at 8k context length. The trick is: a) to use a better quantization algorithm and b) to use unquantized embeddings and the MoE gates (the overhead is quite small). Other notable features include many more…
2024 · its alternatives →
- 14
Last week was big for open source LLMs. We got: - Qwen 2.5 VL (72b and 32b) - Gemma-3 (27b) - DeepSeek-v3-0324 And a couple weeks ago we got the new mistral-ocr model. We updated our OCR benchmark to include the new models. We evaluated 1,000 documents for JSON extraction accuracy. Major takeaways: - Qwen 2.5 VL (72b and 32b) are by far the most impressive. Both landed right around 75% accuracy (equivalent to GPT-4o’s performance). Qwen 72b was only 0.4% above 32b. Within the margin of error. - Both Qwen models passed mistral-ocr (72.2%), which is specifically trained for OCR. - Gemma-3…
2025 · github.com · its alternatives →
- 15
Qwen3.6-35B-A3B▲124The open sparse MoE model for agentic coding
Apr 2026 · qwen.ai · its alternatives →
- 16
Cai▲179Hi HN, Cai runs actions on your selection via ⌥C, pick an action (AI, shell, GitHub issue, Linear ticket), done. Bundled local model (MLX Community), create custom actions with prompts, scripts and more. Free, MIT. Built after my Raycast extension sat in review queue for 3+ months. Wanted no typing, no app switching, no cloud. Repo: https://github.com/cai-layer/cai Demo: https://youtu.be/LH2MYNQ_wsk
Apr 2026 · getcai.app · its alternatives →
- 17
I ported Microsoft's TRELLIS.2 (4B parameter image-to-3D model) to run on Apple Silicon via PyTorch MPS. The original requires CUDA with flash_attn, nvdiffrast, and custom sparse convolution kernels: none of which work on Mac. I replaced the CUDA-specific ops with pure-PyTorch alternatives: a gather-scatter sparse 3D convolution, SDPA attention for sparse transformers, and a Python-based mesh extraction replacing CUDA hashmap operations. Total changes are a few hundred lines across 9 files. Generates ~400K vertex meshes from single photos in about 3.5 minutes on M4 Pro (24GB). Not as fast as…
Apr 2026 · github.com · its alternatives →
- 18
I made this after seeing someone posit the idea online yesterday over lunch then spent some time refining it. So far it's pretty impressive IMO! Right now I am running Qwen3-30B-A3B on my 24gb unified memory m4 MacBook Pro at 50 tok/sec and this should definitely not be working for such a large model on my middling hardware. Things are detailed in the README to get up and running and DESIGN.md has details on all the choices and such made along the way.
Aug 2026 · github.com · its alternatives →
- 19AT
A 3.16M-parameter INT4 transformer running entirely in the on-chip memory of a Xilinx Kria KV260. Zero DRAM in the token loop, 59,965 tok/s on the fabric, bit-exact. Chat with it live.
Aug 2026 · mikeayles.com · its alternatives →
- 20
- 21
Qwen3.8-Flash-Next (MoE, NVFP4) and Qwen3.8-27B on two 16 GB consumer GPUs: two-GPU tensor parallel, experts offloaded to CPU. Formerly ninfer-tp2. - ValerioDolci/capy-flash
8d ago · github.com · its alternatives →
- 22
LTX.bio▲164Hey HN, after our 13B release we've been working on a faster version of our open-source video model and we're excited to share it. We started with a 13B base model that already had competitive generation speeds (e.g. 55s for a 5s video on an H100 — faster than any other model out there). But we wanted to push it further to allow everyone to quickly iterate over their video generations. So we built a distilled version focused on speed without sacrificing temporal or spatial coherence. With the Distilled model, you can now generate 5-second 720p videos in about 9.5 seconds on an H100, and…
2021 · lightricks.com · its alternatives →
- 23WM
Try it out! https://glhf.chat/ Hey HN! We’ve been working for the past few months on a website to let you easily run (almost) any open-source LLM on autoscaling GPU clusters. It’s free for now while we figure out how to price it, but we expect to be cheaper than most GPU offerings since we can run the models multi-tenant. Unlike Together AI, Fireworks, etc, we’ll run any model that the open-source vLLM project supports: we don’t have a hardcoded list. If you want a specific model or finetune, you don’t have to ask us for it: you can just paste the Hugging Face link in and…
2024 · glhf.chat · its alternatives →
- 24MP
Aug 2026 · deepgrove.ai · its alternatives →
Also compare
- Run Qwen3-Next-80B on 8GB GPU at 1tok/2s throughput alternatives
- Run an 80B Qwen in 4.3 GB of RAM on a Mac, and a 35B on an iPhone alternatives
- Qwen3.8-Flash-Next alternatives
- Open-source engine running Gemma 4 26B in 2 GB RAM on any M-series Mac alternatives
- 80% faster, 50% less memory, 0% loss of accuracy Llama finetuning alternatives
- Qwen3.8-Max alternatives
Ranked by how close each launch is in meaning, then by votes. Prices were read from each product’s own site when checked and can change. Refine with your own description →