GLM-5.3-Flash-4bit-paged

Expert-paged build of pipenetwork/GLM-5.3-Flash-MLX-4bit. The weights that are read a fraction at a time live in their own containers, so a machine loads what it needs rather than all of it.

Total 169.27 GiB. Of that, 165.37 GiB is the source build, whose bytes moved into

These containers are not a format mlx-lm reads. The model runs on gbx_lm, a single signed binary for Apple Silicon; there is nothing to pip install. containers rather than being copied, and 3.90 GiB is the draft head, which no published build of this model carries.

Requirements

Mac

Linux (gbx-lm v0.8.1 or later)

On a DGX Spark (128 GB) the model does not fit, so its experts are paged from the SSD.

Memory is not a fixed figure for a paged build, and that is the point of one: it fills what fits and streams the rest from disk. On a 512 GB Mac Studio with room to spare this model settles at about 172 GB resident. A smaller machine holds less and reads more from disk -- slower, but it runs.

How much slower depends on how far the machine is from holding the experts, and on how fast its disk is. Each token routes to a few experts; the ones already in memory cost nothing to reach, and the ones that are not have to be read before that token can finish. A machine holding most of them waits rarely, one holding few waits often. We have not measured this across machine sizes and will not guess a figure: what we can say is that the model answers either way, and that the wait is the SSD's, not the model's.

Install

# upgrading? clear the previous version's unpack directory first
rm -rf ~/.libra/cache/onefile/gbx_lm

# on a DGX Spark, use gbx_lm-linux-arm64.tar.gz in place of gbx_lm-darwin-arm64.tar.gz
curl -fL -o gbx_lm-darwin-arm64.tar.gz 'https://github.com/GreenBitAI/gbx-lm/releases/latest/download/gbx_lm-darwin-arm64.tar.gz' \
  && tar -xzf gbx_lm-darwin-arm64.tar.gz gbx_lm \
  && mkdir -p "$HOME/.local/bin" \
  && mv gbx_lm "$HOME/.local/bin/gbx_lm" \
  && chmod +x "$HOME/.local/bin/gbx_lm"

gbx_lm -h

The Mac build is signed with a Developer ID and notarised, so macOS runs it without the usual detour for a downloaded binary.

command not found -- $HOME/.local/bin is not on your PATH:

echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc && source ~/.zshrc     # zsh
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bash_profile && source ~/.bash_profile   # bash

Killed: 9 (Mac) -- a previous version's files are still in the unpack directory, and macOS refuses to mix two builds. Run the rm -rf line above, then try again.

Run

gbx_lm --model GreenBitAI/GLM-5.3-Flash-4bit-paged

That serves an OpenAI-compatible API on port 11688, which is its default. The weights download on first use into ~/.libra/cache/models; set HF_HOME to put them elsewhere, and HF_TOKEN if you meet the Hub's rate limits for anonymous downloads.

curl http://127.0.0.1:11688/v1/chat/completions \
  -H 'Content-Type: application/json' \
  -d '{"model":"GreenBitAI/GLM-5.3-Flash-4bit-paged","messages":[{"role":"user","content":"Hello"}]}'

Where the weights fit they are filled from experts.bin and the model runs the stock path at stock speed; where they do not, they stream from disk. Reading the machine decides that, not a flag.

To override that: GBX_PAGING=off holds the experts resident.

Checked at build time, while the source checkpoint was still there to compare against:

• PASS layer-wise vs resident — 42 layers x 2 draws exact, 60.93 GiB peak for this gate

Quantization, tokenizer, chat template and licence are unchanged from pipenetwork/GLM-5.3-Flash-MLX-4bit.

Images

The model reads images as well as text. Send them as OpenAI image_url parts -- an http(s) URL or a base64 data: URL:

curl http://127.0.0.1:11688/v1/chat/completions \
  -H 'Content-Type: application/json' \
  -d '{"model":"GreenBitAI/GLM-5.3-Flash-4bit-paged","messages":[{"role":"user","content":[
        {"type":"text","text":"What is in this picture?"},
        {"type":"image_url","image_url":{"url":"https://example.com/photo.jpg"}}]}]}'

From the command line, --images takes one or more files: gbx_lm generate --model GreenBitAI/GLM-5.3-Flash-4bit-paged --images photo.jpg --prompt "What is in this picture?" --max-tokens 2048.

From the command line

From gbx_lm v0.7.1, the binary also runs the model without a server -- one prompt, or an interactive chat:

GBX_GLM53_MTP=on gbx_lm generate --model GreenBitAI/GLM-5.3-Flash-4bit-paged --prompt "Hello" --max-tokens 2048
GBX_GLM53_MTP=on gbx_lm chat --model GreenBitAI/GLM-5.3-Flash-4bit-paged --max-tokens 2048

gbx_lm generate -h and gbx_lm chat -h list their options. Keep --max-tokens generous: the default is 100 for generate and 256 for chat, and the model's reasoning before it answers counts against it. Leave out GBX_GLM53_MTP=on to run without the draft head.

Coding agents

The server speaks three wire protocols on the same port, so the tools that expect a hosted API can be pointed at this one:

Codex -- a provider in ~/.codex/config.toml:

[model_providers.gbx]
name = "gbx-lm"
base_url = "http://127.0.0.1:11688/v1"
wire_api = "responses"

and a profile in ~/.codex/gbx.config.toml:

model_provider = "gbx"
model = "GreenBitAI/GLM-5.3-Flash-4bit-paged"
model_context_window = 1048576

Claude Code -- ~/.claude/gbx.settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "http://127.0.0.1:11688",
    "ANTHROPIC_AUTH_TOKEN": "local",
    "ANTHROPIC_MODEL": "GreenBitAI/GLM-5.3-Flash-4bit-paged",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "GreenBitAI/GLM-5.3-Flash-4bit-paged"
  }
}

Both clients ask for a small model for their own background work, so every name in the settings has to be one this server is serving.

The draft head

The mtp/ folder carries the model's own multi-token prediction head, converted from zai-org/GLM-5.3-Flash, so speculative decoding works from this repository alone. It is off unless asked for:

GBX_GLM53_MTP=on gbx_lm --model GreenBitAI/GLM-5.3-Flash-4bit-paged

Up to 1.4x, more as the context grows. Measured 2026-10-13 with gbx_lm v0.8.0 on a 512 GB Mac Studio (M3 Ultra), weights resident, greedy, thinking off, 256 tokens a reply, decode timed from the first token. Short prompts: three each of code, prose and chat, three runs each, the median per kind. Long contexts: a slice of source code and a question about it, the median of three runs.

Code gains most and prose least. With more context behind it the head proposes better -- it accepts about 63% of what it proposes on short prompts and 75-77% from 14k tokens on.

Every token the head proposes is checked by the model itself, so the reply is the model's own either way; the head only saves passes over the weights.