Qwen3.5-0.8B-decision is an open-weight model from MLX Community, released under Apache License 2.0. It has 873M parameters and a 262,144-token context. At 16-bit it needs about 2.1 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index. It draws 62 downloads a month.
Qwen3.5-0.8B turned into a decision model with optiq lora train --decision: a LoRA on the backbone and a new joint schema head that scores noul, choice and score questions in one pass.
Runs On
What it takes to serve Qwen3.5-0.8B-decision (873M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 1.7 GB | 2.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.9 GB | 1.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.4 GB | 0.5 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 12, 2026.
Qwen3.5-0.8B-decision on every accelerator the SAVRN Index prices, at every precision
Model Card
By MLX Community, published under apache-2.0, revision 36e166ca10c4.
Qwen3.5-0.8B decision model
Qwen3.5-0.8B turned into a decision model with optiq lora train --decision: a LoRA on the backbone and a new joint schema head that scores noul, choice and score questions in one pass. It runs on Apple Silicon through MLX and OptiQ (optiq serve exposes /v1/systemone).
Typed Decisions result
Scored on the full 400-case test split (2,000 decisions) with the benchmark's own metrics.
| Model | Accuracy | KL from gold ↓ | Brier ↓ | ECE ↓ | p50 latency |
|---|---|---|---|---|---|
| This model | 0.772 | 0.097 | 0.054 | 0.177 | 0.34 s |
| gemma-4-e2b-it-OptiQ-4bit-decision | 0.802 | 0.072 | 0.039 | 0.181 | 0.99 s |
| clef-OptiQ-4bit (27B) | 0.719 | 0.189 | 0.102 | 0.118 | 9.7 s |
| Always answer the train-split average | 0.489 | 0.327 | 0.181 | – | – |
Latency is one request per case with all five questions, on an Apple M3 Max.
Configuration
- Architecture
- Qwen3_5ForConditionalGeneration
- Context length (tokens)
- 262,144
- Layers
- 24
- Hidden size
- 1,024
- Feed-forward size
- 3,584
- Attention heads
- 8
- Key/value heads
- 2
- Head dimension
- 256
- Vocabulary size
- 248,320
- Model type
- qwen3_5
Identity and Version
- Repository
- mlx-community/Qwen3.5-0.8B-decision
- Publisher
- MLX Community
- Task
- Not stated by the source
- Modality
- Other
- Library
- mlx
- Parameters
- 873M parameters
- Languages
- mlx
- Revision
- 36e166ca10c4d16471a231d66cfc22d80039fe34
- First published
- 2026-10-08
- Last updated
- 2026-10-11
Files and Weights
19 files, 1.9 GB in total. The weights are 3 files totalling 1.8 GB in safetensors.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| adapters.safetensors | Weights | 40.9 MB | 7ad3e00365ba |
| joint_head.safetensors | Weights | 54.7 MB | 2ab266645287 |
| model.safetensors-00001-of-00001.safetensors | Weights | 1.7 GB | 04b1c301231d |
| .eval_results/typed-decisions.yaml | Configuration | 2.1 KB | — |
| adapter_config.json | Configuration | 479 B | — |
| config.json | Configuration | 2.9 KB | — |
| decision_config.json | Configuration | 97 B | — |
| joint_head_config.json | Configuration | 116 B | — |
| model.safetensors.index.json | Configuration | 50.9 KB | — |
| preprocessor_config.json | Configuration | 390 B | — |
| video_preprocessor_config.json | Configuration | 385 B | — |
| LICENSE | Documentation | 11.5 KB | — |
| README.md | Documentation | 2.0 KB | — |
| chat_template.jinja | Other | 7.8 KB | — |
| .gitattributes | Repository | 1.6 KB | — |
| merges.txt | Tokenizer | 3.4 MB | — |
| tokenizer.json | Tokenizer | 12.8 MB | 5f9e4d4901a9 |
| tokenizer_config.json | Tokenizer | 16.7 KB | — |
| vocab.json | Tokenizer | 6.7 MB | — |
License and Download
- License
- apache-2.0
- Access
- Open weights, no gate
- Download size
- 1.8 GB
Released by MLX Community through its official repository on Hugging Face. Read the license.
Built From
- Derived from Qwen/Qwen3.5-0.8B
Evaluations
Each result is shown as reported, with the conditions its reporter stated. None is a SAVRN measurement. A comparison lines two results up only when their configuration, unit and setup are all stated and identical.
| Benchmark | Conditions | Result | Reported by | Revision | Date |
|---|---|---|---|---|---|
| LocalLLaMA/typed-decisions | Task accuracyMetric accuracySetup Fitted by us on the benchmark's own train split only (1,080 records, LoRA on the backbone plus a new schema head), then scored on the full 400-case test split (2,000 decisions): one request per case with all five questions, text only, the benchmark's own metrics, MLX on an Apple M3 Max.Comparison conditions not established | 0.772 | Typed Decisions leaderboard (README) Reported by a third party |
Evaluated revision not stated | 2026-10-11 |
| LocalLLaMA/typed-decisions | Task brierMetric brierSetup Fitted by us on the benchmark's own train split only (1,080 records, LoRA on the backbone plus a new schema head), then scored on the full 400-case test split (2,000 decisions): one request per case with all five questions, text only, the benchmark's own metrics, MLX on an Apple M3 Max.Comparison conditions not established | 0.054 | Typed Decisions leaderboard (README) Reported by a third party |
Evaluated revision not stated | 2026-10-11 |
| LocalLLaMA/typed-decisions | Task eceMetric eceSetup Fitted by us on the benchmark's own train split only (1,080 records, LoRA on the backbone plus a new schema head), then scored on the full 400-case test split (2,000 decisions): one request per case with all five questions, text only, the benchmark's own metrics, MLX on an Apple M3 Max.Comparison conditions not established | 0.1769 | Typed Decisions leaderboard (README) Reported by a third party |
Evaluated revision not stated | 2026-10-11 |
| LocalLLaMA/typed-decisions | Task kl_from_goldMetric kl_from_goldSetup Fitted by us on the benchmark's own train split only (1,080 records, LoRA on the backbone plus a new schema head), then scored on the full 400-case test split (2,000 decisions): one request per case with all five questions, text only, the benchmark's own metrics, MLX on an Apple M3 Max.Comparison conditions not established | 0.0974 | Typed Decisions leaderboard (README) Reported by a third party |
Evaluated revision not stated | 2026-10-11 |
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 1.8 GB |
| 16-bit | 1.7 GB |
| 8-bit | 0.9 GB |
| 4-bit | 0.4 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About Qwen3.5-0.8B-decision
How much GPU memory does Qwen3.5-0.8B-decision need?
About 2.1 GB at 16-bit and 0.5 GB at 4-bit: the weights (873M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run Qwen3.5-0.8B-decision on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use Qwen3.5-0.8B-decision commercially?
Yes. Qwen3.5-0.8B-decision is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.
What is Qwen3.5-0.8B-decision's context length?
262,144 tokens, from the maximum position embeddings in its published configuration.