GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
cagataydev
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
cagataydev
Model Tree
Qwen/Qwen3.5-0.8B
Input Modalities
Output Modalities
Supported Functionality
Agent(model:tools:) and a small, honest loop: model → tools → model until it answers. Tool errors go
back to the model (it fixes the call instead of repeating it); anything that leaves the device waits for a tap.
Runtimes for MLX, llama.cpp and Apple's on-device model; hooks, checkpoints, structured output, tracing.loopl.dev · Get it on TestFlight · github.com/cagataycali/loopl
| probe | this model | previous (v1) | base Qwen3.5 4-bit |
|---|---|---|---|
| agent-SDK knowledge quiz (first 472 questions, exact-answer) | 74 | 59 | 50 |
| tool calls, native template dialect | 9/9 | 7/9 | 6/9 |
| tool calls, loopl JSON dialect | 7/9 | 7/9 | 2/9 |
| identity (5 classic probes, with system prompt) | 5/5 | 4/5 | 4/5 |
| identity · bio facts (8, with system prompt) | 8/8 | — | — |
| identity · (13) |
Scored by train/eval.py on the 4-bit MLX export (greedy). The quiz asks for exact file-level facts about an agent SDK's source — hard for every small model; the delta over base is the point. Tool probes count a well-formed call with the right arguments. identity_nosys is the one that matters in the app: the system prompt there is the user's to edit.
Vision: the base's tower is frozen and kept — 153 vision_tower.* tensors in the MLX export, so the app shows the photo button.
In the loopl app (iPhone). Install from TestFlight → Models → the loopl models section lists this size (no account, no token) → download. The app sees the vision tower and shows the photo button; the system prompt is yours to edit — identity and tool manners are in the weights.
On a Mac with MLX (vision included):
bash
pip install -U mlx-vlmpython -m mlx_vlm.generate --model cagataydev/loopl-0.8b-4bit \ --image photo.jpg --prompt "What is in this picture? Then tell me who built you." --max-tokens 200With llama.cpp: the GGUF Q4_K_M export (cagataydev/loopl-0.8b-GGUF) is not public yet; export your own from the bf16 weights with
llama.cpp's convert_hf_to_gguf.py --no-mtp (text only — the converter has no vision path).
From Swift with the loopl SDK — the snippet below is docs/start.md § "Run the loop" in the loopl repo,
compile-checked in CI against the package (products Loopl + LooplRuntimes); spec is this repo's ModelSpec and dir the
downloaded folder:
swift
import Looplimport LooplRuntimesimport Foundation func chat(_ spec: ModelSpec, at dir: URL) async throws { let model = MLXModel() // Metal — a real device try await model.load(spec, from: dir) { _ in } let agent = Agent(model: model, tools: [CurrentTimeTool(), CalculatorTool()], systemPrompt: "You are a concise assistant on the user's phone.") for try await event in agent.stream("What time is it in Istanbul, and what is 23 × 47?") { switch event { case .text(let t): print(t, terminator: "") case .toolStarted(let use): print("\n→ \(use.name)") case .toolResult(let r): print("← \(r.content)") default: break } }}Keep post-tuning it. The bf16 weights (cagataydev/loopl-0.8b) are the base for continual post-tuning: in the app, Models › Train your own takes a loopl model as the base and trains on your own shared conversations (
text
train/loopl_sft.py --base cagataydev/loopl-0.8btransformers ≥ 5 as Qwen3_5ForConditionalGeneration. Export your own
quantisation with mlx_vlm.convert -q (keeps vision) — mlx_lm.convert silently drops the tower.
| base | Qwen/Qwen3.5-0.8B (vision tower frozen, visual.* never trained) |
| data | cagataydev/loopl-train (private) @ 47d27b44 · 13704 rendered rows / 37,226,229 tokens · both tool dialects · oversample c=2 |
| method | full fine-tune of the language model · lr 2e-05 · batch 2×8 · max_len 4096 · 2.0 epochs, best-epoch checkpoint kept |
| eval loss | ep1 1.856, ep2 1.910 → best 1.856 (eval split of the same dataset revision) |
| compute |
Script: train/loopl_sft.py — the same single file the loopl app launches when you
tap Models › Train your own on your own conversations. Assistant turns that seed a bad tool call carry weight: 0
so the model learns the recovery, not the mistake. Tokenizer files are normalised to the base's (Qwen2Tokenizer,
chat_template.jinja with tools) so Swift loaders accept them; train/check_mlx_repo.py gates every export on that
plus the vision triad (vision_config ⇔ vision_tower.* ⇔ preprocessor_config.json).
calculator, http, recall) and it does better.Every training row is agent-synthesised from public sources (the owner's public GitHub repos and the agent-SDK source they build on), critic-reviewed (score ≥ 4/5 kept) and filtered by a denylist for secrets, private names, phone numbers and addresses — see the dataset card. No user conversations from the app are in this model.
Apache-2.0, inherited from the Qwen3.5 base. The fine-tune, exports and dataset are © Cagatay Cali, same license.
Six public repos, one recipe. The app lists the three MLX rows under Models › loopl models (2B is the sweet spot for speed on an
iPhone 15/16; 4B is the most knowledgeable; 0.8B fits anywhere). Every bf16 repo is a valid --base for the next round.
| MLX 4-bit · vision · what the app downloads | bf16 · the base to keep training | |
|---|---|---|
| 0.8B | cagataydev/loopl-0.8b-4bit · 625 MB | cagataydev/loopl-0.8b · 1.71 GB ← this repo |
| 2B | cagataydev/loopl-2b-4bit · 1.72 GB | · 4.43 GB |
All of them: the loopl collection.
| 12/13 |
| — |
| — |
a100-large (Hugging Face Jobs), 9143 s train · job 6ac5e6e2404719ba37665442 |
| stack | transformers 5.19.0 · trl 1.14.2 · peft 0.21.2 |
| exports | MLX 4-bit via mlx_vlm.convert -q (153 vision tensors of 847, 625 MB) · GGUF Q4_K_M via llama.cpp --no-mtp (529 MB, text only) |
cagataydev/loopl-2b| 4B | cagataydev/loopl-4b-4bit · 3.03 GB | cagataydev/loopl-4b · 9.08 GB |