GLM-5-FP8
Support summary
GLM-5-FP8 is a checkpoint published by Z.ai, built on the GlmMoeDsaForCausalLM architecture. Official recipes or cookbooks cover vLLM and SGLang, without pinning a stable release.
- Documented quantization
- FP8
- Documented features
- Multi-token prediction, Reasoning parsers, Speculative decoding, Tensor parallelism, and Tool calling
- Configured context
- 202,752 tokens
- License
- mit
Hardware, quantization and features count only statements about this checkpoint; engine-wide capabilities are listed with each release.
Hugging Face model card
- Publisher
- Z.ai
- Task
- text generation
- License
- mit
- Library
- transformers
- Architecture
- GlmMoeDsaForCausalLM
- Weight tensor elements
- 753.9B
- Tensor types
- F32, BF16, F8_E4M3
- Quantization
- fp8
- Configured context
- 202,752 tokens
- Languages
- en, zh
- Weight access
- Public
- Repository created
- 2026-02-11
Publisher deployment guidance
- vLLM>=0.19.0
Publisher minimum version for this checkpoint; follow its launch recipe and backend requirements.
Source: zai-org/GLM-5-FP8 · publisher model cardReviewed - SGLang>=0.5.10
Publisher minimum version for this checkpoint; follow its launch recipe and backend requirements.
Source: zai-org/GLM-5-FP8 · publisher model cardReviewed
Engine support
3 engines
| Engine | Release | Checkpoint support | Evidence scope | Last known supported |
|---|---|---|---|---|
| vLLM | Version not pinnedRecipe / cookbook | Documented recipeSource: vllm · models/zai-org/GLM-5.yamlReviewed | No supported release indexed | |
Features, dependencies and evidence for vLLM 0.28.0 | ||||
| SGLang | Version not pinnedRecipe / cookbook | Documented recipeSource 1: sglang · docs/cookbook/autoregressive/GLM/GLM-5.mdxSource 2: sglang · docs/src/snippets/autoregressive/glm-5-deployment.jsxReviewed | No supported release indexed | |
Features, dependencies and evidence for SGLang 0.5.19 | ||||
| TensorRT-LLM | 1.2.1Newest indexed | No matching evidence curated | No supported release indexed | |
Features, dependencies and evidence for TensorRT-LLM 1.2.1 | ||||