GLM-5.3

The full-size sibling of GLM-5.3 Flash and the second-ranked open-weight model on aggregate benchmarks as of September 2026. Its parameter count and MLA geometry are taken from revision-pinned official weights and configuration metadata.

Z.aiMixture of expertscodingreasoning2 machines can run it

Run locally →

Total parameters

753.33B

Official spec sets memory need

Active parameters

40B

Official spec sets decode speed

Context

Official spec tokens

4-bit weights

410 GB

Calculated before KV cache

This is a sparse mixture of experts. All 753.33B of weights must be available to the runtime, but only 40B are read per generated token. Ordinary runtimes keep the full quantized model in memory; a runtime-specific SSD-streaming artifact can retain a smaller working set and fetch expert data on demand, trading speed for capacity.

Model specification

Publisher

Z.ai Official spec

Architecture

Sparse mixture of experts Official spec

Total parameters

753.33B Official spec

Active parameters

40B per token Official spec

Context length

1,048,576 tokens Official spec

Attention ⓘ

MLA · 64 KV heads × 192 dim × 78 layers Official spec

Licence

MIT Official spec source

Specification confidence ⓘ

High · verified 2026-10-19 Official spec source

Released

1 Jul 2026 Official spec

Official source

Hugging Face

Quantizations and memory

Quantization	Format	Bits/weight	Resident	Download	Quality kept
`MLX 4-bit`	mlx	4.5	410.4 GB	410.4 GB	98.0%
`Q4_K_M`default	gguf	4.85	438.1 GB	438.1 GB	98.5%
`Q5_K_M`	gguf	5.69	514.0 GB	514.0 GB	99.3%
`Q8_0`	gguf	8.5	760.4 GB	760.4 GB	99.9%

Generic weight sizes are computed from the parameter count and bits per weight plus a format-specific overhead. Runtime-specific artifacts use their published resident and download footprints; streamed models can therefore require much more disk than memory. Quality retention is an assumption, not a measured evaluation.

Coding & quality benchmarksCompare coding results →

Benchmark	Category	Score	Run details	Reported by	Date	Source
Terminal-Bench 2.1	agentic	88.2	Not supplied by source	canonical model_repository	31 Aug 2026	link
DeepSWE	coding	66.9	Not supplied by source	canonical model_repository	28 Aug 2026	link

We show source metrics rather than deriving one opaque quality number. Different benchmarks measure genuinely different things, and collapsing them into a single score would hide exactly the disagreements worth seeing.

Recommendations

Cheapest that can run it

NVIDIA DGX Station GB300 (748 GB)

~29.5–42.5 t/s · $90,082

Lowest purchase price among configurations where the model fits at some quantization in our catalogue. Speed is not considered.

Cheapest above 20 t/s

NVIDIA DGX Station GB300 (748 GB)

~29.5–42.5 t/s · $90,082

20 tokens/second is roughly the point at which generation keeps pace with reading. Below it, interactive use feels like waiting.

Cheapest above 40 t/s

NVIDIA DGX H200 (8x H200, 1,128 GB)

~254–365 t/s · $359,430

40 tokens/second is the threshold most people describe as comfortable for coding agents, where output arrives faster than you can review it.

Fastest with real measurements

Nothing in the database qualifies.

Highest throughput among configurations with an actual published measurement rather than our estimate.

Best throughput per purchase-price unit

NVIDIA DGX H200 (8x H200, 1,128 GB)

~254–365 t/s · $359,430

Highest decode tokens/second per 1,000 units of the displayed purchase currency. Ignores running costs and resale — see the economics section for the full picture.

Hardware that runs GLM-5.30 measured, 2 estimated

Quantization

Minimum decode: 0 t/s

2 of 2 rows

Hardware↕	Quant	Memory↕	Decode▼	Prefill↕	Context	Price↕	Fit	Confidence↕
NVIDIA DGX H200 (8x H200, 1,128 GB) NVIDIA · 1,128 GB · 38,400 GB/s	`Q8_0` recommended	787.6 GB	~254–365 t/s	~15410–32010 t/s	1024K	$359,430	Comfortable	Estimated
NVIDIA DGX Station GB300 (748 GB) NVIDIA · 748 GB · 7,100 GB/s	`Q4_K_M` recommended	452.9 GB	~29.5–42.5 t/s	~2060–4280 t/s	1024K	$90,082	Comfortable	Estimated

Submit a benchmarkContributions are reviewed before publication.

Hosted alternatives for this exact model

These endpoints serve the same open weights, so comparing them against local ownership is a like-for-like economic question rather than a quality trade-off.