models · zai-org

GLM-5.3-Flash

text-generation model harvested from Hugging Face (trending). Catalogued at Quarantine — unverified, pending attestation.

Compare Verify

What this registry has checked

Quarantine. We hold the publisher’s own published hash for these weights and have recorded it against this entry. We have not downloaded and hashed the weights ourselves, so the hash is the publisher’s claim, not our measurement.

Method	hf-published-sha256

Re-check this yourself → Source → Full record →

What is recorded

Publisher	zai-org
Origin	China
Licence	mit
Parameters	321,323,031,390 (321.3B)
Context window	1,048,576 tokens
Modality	text+image+video->text
Native precision	F8_E4M3

Reported scores

GPQA Diamond	91.2
Humanity's Last Exam	39.9

Reported by a third party and recorded here, not re-run by us. Source: artificialanalysis.ai/api/v2.

Measured on real hardware

None of these runs are ours. Each figure is the reporter’s, linked to their own evidence, under their own trust label. The number shown is single-stream decode — one user, concurrency 1 — because aggregate throughput across many users is a different quantity and reads as far faster than anyone will see.

Reported to local-ai-registry

Hardware	Engine	Build measured	Decode, 1 stream	Label	Reported by
NVIDIA DGX Spark GB10 ×4 devices	vllm	ModelOpt NVFP4	112 tok/s @ 8,182	candidate	0xsero
Apple M3 Ultra 512GB	vllm-mlx-native	MLX 4bit	41.3 tok/s @ 6,638	candidate	native-studio-measurement
Apple M3 Ultra 512GB	vllm-mlx-native	MLX 4bit	34.5 tok/s @ 8,192	candidate	native-studio-measurement
Radeon AI PRO R9700	llama.cpp	UD-IQ1_M	17.4 tok/s @ 32,768	candidate	localmaxxing
Radeon AI PRO R9700	llama.cpp	UD-Q2_K_XL	16.6 tok/s @ 32,768	candidate	localmaxxing
Measured outside 2K–32K context — decode speed depends heavily on context, so these are not comparable with the rows above
RTX PRO 6000 Blackwell ×4 devices	vllm	NVFP4 NVFP4 experts / BF16 sensitive target weights	404 tok/s @ 971	validated	0xsero
RTX PRO 6000 Blackwell ×4 devices	vllm	safetensors EXL3 4bpw	185 tok/s	candidate	0xsero
RTX PRO 6000 Blackwell ×2 devices	vllm	safetensors EXL3 4bpw	145 tok/s	candidate	0xsero
RTX PRO 6000 Blackwell ×4 devices	sglang	ModelOpt NVFP4	142 tok/s	candidate	0xsero
RTX PRO 6000 Blackwell ×4 devices	sglang	selective EXL3 TP4 safetensors 4.0 bpw routed experts with BF16 backbone	73.6 tok/s @ 126	candidate	0xsero

From local-ai-registry (MIT), commit 124959d: 44 runs across 5 machines, 3 validated — model revision and runtime pinned, launch accepted — and 41 candidate, their label for useful evidence without a reproducible-launch promise. 6 more not listed.

Elsewhere in the registry

All models → Browse with filters → What will run on my hardware? →

Registry record as of 2026-09-01. Identity, licence and parameter count are properties of a release and do not move; tier can, and the verify link above re-checks it against the live registry rather than this page.