# GLM-5.3-Flash: official model and deployment card

Official Flash checkpoint and local-serving references. Its reasoning effort and chat-template controls matter when attempting evaluation reproduction.

- Type: Official model card
- Evidence: source_reported; site reproduced: false
- Model family: GLM
- Model: zai-org/GLM-5.3-Flash
- Original author: zai-org
- Original project: https://huggingface.co/zai-org/GLM-5.3-Flash
- Pinned document: https://huggingface.co/zai-org/GLM-5.3-Flash/blob/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/README.md
- Source revision: 46a050bec13002ad29ee795c16e61cf4fa3c4a9c
- Source commit time (not experiment date): 2026-10-15T06:32:52.000Z
- Checked: 2026-10-17
- Experiment date: None
- Runtime topics: vLLM, SGLang, Transformers, KTransformers, TokenSpeed
- Hardware topics: 
- Reported setup: {"system": null, "node_count": null, "parallelism": null, "kv_dtype": null, "context_limit_tokens": null, "experiment_date": null}
- Performance figures: attributed selections with per-point conditions; no cross-source ranking

## Conditions

Unknown is null. Recipe options are not measured benchmark settings.

```json
{
  "conditions": {
    "model_revision": "46a050bec13002ad29ee795c16e61cf4fa3c4a9c",
    "weight_hash": null,
    "runtime_execution_revision": null,
    "input_tokens": null,
    "output_tokens": null,
    "concurrency": null,
    "cache_state": null,
    "quantization": null,
    "speculative_decoding": null,
    "temperature": null,
    "top_p": null
  },
  "test_profiles": [],
  "recipe_parameters": null,
  "deployment_options": []
}
```

## Reading notes

- The card specifies low/high/max reasoning effort and a default of max.
- The clear_thinking chat setting is a separate control.

## Limits

- Flash and the full GLM-5.3 checkpoint have separate artifacts and terms.

## License declaration

{"declared_spdx": "mit", "declared_name": null, "scope": "Publisher model-card declaration; inspect the linked LICENSE and artifact-specific terms."}


## Editorial answer

Official Flash checkpoint and local-serving references. Its reasoning effort and chat-template controls matter when attempting evaluation reproduction.

By Run AI Here editorial · 2026-10-17

AI-assisted source reading and original editorial synthesis; no independent model execution.

## What distinguishes Flash from the full GLM model?

The Flash card describes a newly trained, natively multimodal model with 320B total and 18B active parameters, using sparse and linear attention with mHC. The deployment recipe rounds the total differently at 321B. This page retains the publisher’s wording and the recipe’s number in their own records; neither establishes memory use under a serving workload.

Sources: https://huggingface.co/zai-org/GLM-5.3-Flash/blob/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/README.md#L25

## Which reasoning controls affect reproduction?

reasoning_effort accepts low, high and max. Omitting it, or supplying another value, selects max; the publisher asks benchmark reproductions to retain that setting. clear_thinking is independent and defaults to false. The card recommends explicitly enabling clear_thinking for chat. Keep both settings in an evaluation record rather than inferring one from the other.

Sources: https://huggingface.co/zai-org/GLM-5.3-Flash/blob/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/README.md#L45

## Why are benchmark context budgets different?

HLE with tools uses a 300,000-token managed context and a 163,840-token generation ceiling, with temperature 1.0 and top_p 0.95. DeepSWE instead specifies 400K context, temperature 0.95, top_p 1.0 and a six-hour timeout. These are task-specific harness settings, not one global chat preset or a local speed measurement.

Sources: https://huggingface.co/zai-org/GLM-5.3-Flash/blob/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/README.md#L50

## Where should a deployment begin?

The card links model-specific vLLM, SGLang, TokenSpeed, Transformers and other integration guides. Choose a versioned engine route, then match the actual artifact and hardware profile. A link in the publisher card establishes an intended integration path; this collection has not installed those stacks or verified their performance on a reader’s device.

Sources: https://huggingface.co/zai-org/GLM-5.3-Flash/blob/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/README.md#L32

## Attributed configuration facts

```json
[
  {
    "name": "Total / active parameters as counted by card",
    "value": "320B / 18B",
    "classification": "publisher_declaration",
    "evidence_url": "https://huggingface.co/zai-org/GLM-5.3-Flash/blob/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/README.md#L25",
    "source_document_sha256": "e6c5d41400e28f5a89418a0d15b09b9c113314a7553d40d7d4a8a72fcdb87d02"
  },
  {
    "name": "Reasoning effort",
    "value": "low / high / max; default max",
    "classification": "publisher_declaration",
    "evidence_url": "https://huggingface.co/zai-org/GLM-5.3-Flash/blob/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/README.md#L45",
    "source_document_sha256": "e6c5d41400e28f5a89418a0d15b09b9c113314a7553d40d7d4a8a72fcdb87d02"
  },
  {
    "name": "Chat thinking cleanup",
    "value": "clear_thinking false by default; true recommended for chat",
    "classification": "publisher_declaration",
    "evidence_url": "https://huggingface.co/zai-org/GLM-5.3-Flash/blob/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/README.md#L46",
    "source_document_sha256": "e6c5d41400e28f5a89418a0d15b09b9c113314a7553d40d7d4a8a72fcdb87d02"
  },
  {
    "name": "HLE-with-tools context / output cap",
    "value": "300000 / 163840 tokens",
    "classification": "publisher_declaration",
    "evidence_url": "https://huggingface.co/zai-org/GLM-5.3-Flash/blob/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/README.md#L50",
    "source_document_sha256": "e6c5d41400e28f5a89418a0d15b09b9c113314a7553d40d7d4a8a72fcdb87d02"
  }
]
```