Page values for "GLM 5.3 Flash on 3x 8 GB"
From CMP Wiki
More actions
"Recipe" values
1 row is stored for this page| Field | Field type | Allowed values | Value |
|---|---|---|---|
| page | Page | GLM 5.3 Flash on 3x 8 GB | |
| model | String | GLM-5.3-Flash (320B MoE, 18B active) | |
| quant | String | W4A16, about 4.3 bpw, 163 GiB on disk | |
| engine | String | Morrowmake/vllm-cmp170hx (via the glm53-flash-cmp170hx-recipe) | |
| engine_version | String | not stated | |
| cards | String | 8 GB · 10 GB · both | 8 GB |
| card_count | Integer | 3 | |
| topology | String | PP3 | |
| context | String | about 1,000,000 (8-bit KV cache) | |
| decode_tps | Float | ||
| prefill_tps | Float | ||
| vram_per_card | String | 64 GiB | |
| power_cap | String | ||
| driver | String | ||
| unlock_build | String | ||
| flags | Text | Morrowmake's recipe with pipeline parallel 3; KV cache quantized to 8 bits | |
| status | String | Proven · Reported · Theory · Contested · Warning | Reported |
| tested_by | String | shirogeek | |
| tested_on | Date | 2026-10-19 | |
| source | String | https://discord.com/channels/1518634379604136178/1518695289278566450/1561743826945843304 | |
| files | Wikitext |