Beam 1.0 501B A23B vs GLM 5.2
GLM 5.2 leads llmrun Score 122.9 to 119.5 (ranges overlap); neither fits 24 GB.
Model A · reflection
Beam 1.0 501B A23B501.3B · 23.4B active · 1M context · Apache 2.0 · Oct 2026
Model B · Z.ai
GLM 5.2753.3B · 41.8B active · 1M context · MIT · Jun 2026
Choose Beam 1.0 501B A23B if you want…
Lighter on memory
301.4 vs 456.2 GB at Q4_K_M
Also does reasoning, code
GLM 5.2 doesn't list it
Newer release
Oct 2026 vs Jun 2026
Choose GLM 5.2 if you want…
Higher llmrun Score
122.9 vs 119.5, ranges overlap
Stronger at agents
128.1 vs 118.3
On your hardware
Each model at its default quant (Q4_K_M), a smaller one when it must. Estimates, ±20%.
8 GB
RTX 4060 · 8 GB tier
Beam 1.0 501B A23B
Doesn't fit
smallest quant needs 141.1 GB
GLM 5.2
Doesn't fit
smallest quant needs 208.9 GB
12 GB
RTX 3060 12GB · 12 GB tier
Beam 1.0 501B A23B
Doesn't fit
smallest quant needs 141.1 GB
GLM 5.2
Doesn't fit
smallest quant needs 208.9 GB
16 GB
RTX 5060 Ti 16GB · 16 GB tier
Beam 1.0 501B A23B
Doesn't fit
smallest quant needs 141.1 GB
GLM 5.2
Doesn't fit
smallest quant needs 208.9 GB
24 GBreference
RTX 4090 · 24 GB tier
Beam 1.0 501B A23B
Doesn't fit
smallest quant needs 141.1 GB
GLM 5.2
Doesn't fit
smallest quant needs 208.9 GB
32 GB
RTX 5090 · 32 GB tier
Beam 1.0 501B A23B
Doesn't fit
smallest quant needs 141.1 GB
GLM 5.2
Doesn't fit
smallest quant needs 208.9 GB
48 GB
RTX A6000 · 48 GB tier
Beam 1.0 501B A23B
Doesn't fit
smallest quant needs 141.1 GB
GLM 5.2
Doesn't fit
smallest quant needs 208.9 GB
Mac 32 GB
Mac Mini M4 · Macs
Beam 1.0 501B A23B
Doesn't fit
smallest quant needs 141.1 GB
GLM 5.2
Doesn't fit
smallest quant needs 208.9 GB
Mac 64 GB
Mac Studio M4 Max · Macs
Beam 1.0 501B A23B
Doesn't fit
smallest quant needs 141.1 GB
GLM 5.2
Doesn't fit
smallest quant needs 208.9 GB
Mac 128 GB
Mac Studio M4 Max · Macs
Beam 1.0 501B A23B
Doesn't fit
smallest quant needs 141.1 GB
GLM 5.2
Doesn't fit
smallest quant needs 208.9 GB
Reference GPU: RTX 4090. Tinted cell = the better pick on that machine (fits, more headroom, or at least 10% faster).
Overall
Beam 1.0 501B A23B
GLM 5.2
Coding
Beam 1.0 501B A23B
no data
GLM 5.2
Agents
Beam 1.0 501B A23B
GLM 5.2
Math
Beam 1.0 501B A23B
no data
GLM 5.2
Science
Beam 1.0 501B A23B
no data
GLM 5.2
Reasoning
Beam 1.0 501B A23B
GLM 5.2
Beam 1.0 501B A23B · #17 of 106 open models · #14 of 102 that fit ≤512 GB
GLM 5.2 · #13 of 106 open models · #10 of 102 that fit ≤512 GB
Benchmarks
4 shared boards: Beam 1.0 501B A23B leads 2, GLM 5.2 leads 2 · Benchmarks imported Nov 2026
Beam 1.0 501B A23B
23.5%#11/22 open
IndependentEpoch AI · Oct 2026
GLM 5.2
22.8%#12/22 open
IndependentEpoch AI · Jun 2026
Beam 1.0 501B A23B
74.0%#12/22 open
IndependentEpoch AI · Oct 2026
GLM 5.2
77.0%#11/22 open
IndependentEpoch AI · Jun 2026
Beam 1.0 501B A23B
48.8%#6/20 open
Lab-reportedEpoch AI · Oct 2026
GLM 5.2
45.2%#10/20 open
Lab-reportedEpoch AI · Jun 2026
Beam 1.0 501B A23B
18.0%#16/31 open
Lab-reportedEpoch AI · Oct 2026
GLM 5.2
35.0%#11/31 open
Lab-reportedEpoch AI · Jun 2026
32 more boards (other shared and single-model boards)
Beam 1.0 501B A23B
1435.7#23/188 open
ArenaLMArena · Oct 2026
GLM 5.2
1475.3#6/188 open
ArenaLMArena · Oct 2026
Beam 1.0 501B A23B
1436.0#25/55 open
ArenaLMArena · Oct 2026
GLM 5.2
1605.6#10/55 open
ArenaLMArena · Oct 2026
Only Beam 1.0 501B A23B (7)
Only GLM 5.2 (23)
- GPQA Diamond91.9%
- LiveBench Coding79.7
- LiveBench Reasoning78.6
- AIME 2024/202586.4%
- FrontierMath59.2%
- LiveBench Math89.8
- SciCode51.2%
- SimpleQA34.2%
- SimpleBench58.8%
- FrontierMath Tier 429.3%
- WeirdML70.1%
- DeepSWE43.8%
- FrontierCode24.5%
- Chess Puzzles22.0%
- CritPt20.9%
- DTBench93.6%
- EBR-bench9.5%
- LMCA45.8%
- Mystery Game Puzzles19.0%
- Vending-Bench 28313.8
- ALE-Bench1047.0
- Surface Evolver Bench55.6%
- GBA Eval0.0%
VRAM by quantization
Memory to load the model and 2K context; below it, the total at 32K and 128K context. Lighter side in bold.
Q2_K
Beam 1.0 501B A23B
214 GB190.7 GB file
32K 219 GB · 128K 237 GB
GLM 5.2
324 GB~320.2 GB file
32K 323 GB · 128K 332 GB
Q3_K_S
Beam 1.0 501B A23B
220 GB225.9 GB file
32K 225 GB · 128K 243 GB
GLM 5.2
not listed
Q3_K_M
Beam 1.0 501B A23B
245 GB236.6 GB file
32K 250 GB · 128K 268 GB
GLM 5.2
371 GB342.7 GB file
32K 370 GB · 128K 379 GB
Q4_0
Beam 1.0 501B A23B
251 GB284.2 GB file
32K 257 GB · 128K 274 GB
GLM 5.2
not listed
Q4_K_M
Beam 1.0 501B A23B
301 GB314.5 GB file
32K 307 GB · 128K 324 GB
GLM 5.2
456 GB465.8 GB file
32K 455 GB · 128K 464 GB
Q5_K_M
Beam 1.0 501B A23B
358 GB380.8 GB file
32K 363 GB · 128K 381 GB
GLM 5.2
541 GB560.8 GB file
32K 540 GB · 128K 549 GB
Q6_K
Beam 1.0 501B A23B
414 GB441.8 GB file
32K 420 GB · 128K 437 GB
GLM 5.2
626 GB625.9 GB file
32K 625 GB · 128K 634 GB
Q8_0
Beam 1.0 501B A23B
502 GB532.8 GB file
32K 507 GB · 128K 525 GB
GLM 5.2
758 GB801.4 GB file
32K 757 GB · 128K 765 GB
15 more quants
IQ2_XXS
Beam 1.0 501B A23B
139 GB145.3 GB file
32K 144 GB · 128K 161 GB
GLM 5.2
211 GB238.5 GB file
32K 210 GB · 128K 219 GB
IQ2_XS
Beam 1.0 501B A23B
151 GB151.7 GB file
32K 156 GB · 128K 174 GB
GLM 5.2
230 GB230.0 GB file
32K 229 GB · 128K 238 GB
IQ2_S
Beam 1.0 501B A23B
157 GB161.3 GB file
32K 163 GB · 128K 180 GB
GLM 5.2
not listed
IQ2_M
Beam 1.0 501B A23B
170 GB178.5 GB file
32K 175 GB · 128K 193 GB
GLM 5.2
258 GB238.6 GB file
32K 257 GB · 128K 266 GB
IQ3_XXS
Beam 1.0 501B A23B
195 GB211.1 GB file
32K 200 GB · 128K 218 GB
GLM 5.2
296 GB281.7 GB file
32K 295 GB · 128K 304 GB
IQ3_XS
Beam 1.0 501B A23B
207 GB224.5 GB file
32K 213 GB · 128K 230 GB
GLM 5.2
not listed
IQ3_S
Beam 1.0 501B A23B
not listed
GLM 5.2
324 GB308.6 GB file
32K 323 GB · 128K 332 GB
IQ3_M
Beam 1.0 501B A23B
226 GB255.7 GB file
32K 232 GB · 128K 249 GB
GLM 5.2
343 GB356.3 GB file
32K 342 GB · 128K 351 GB
Q3_K_L
Beam 1.0 501B A23B
258 GB249.5 GB file
32K 263 GB · 128K 280 GB
GLM 5.2
not listed
IQ4_XS
Beam 1.0 501B A23B
270 GB273.1 GB file
32K 276 GB · 128K 293 GB
GLM 5.2
409 GB365.3 GB file
32K 408 GB · 128K 417 GB
IQ4_NL
Beam 1.0 501B A23B
283 GB311.9 GB file
32K 288 GB · 128K 306 GB
GLM 5.2
428 GB372.7 GB file
32K 427 GB · 128K 436 GB
Q4_1
Beam 1.0 501B A23B
283 GB314.1 GB file
32K 288 GB · 128K 306 GB
GLM 5.2
not listed
Q4_K_S
Beam 1.0 501B A23B
283 GB293.4 GB file
32K 288 GB · 128K 306 GB
GLM 5.2
428 GB436.4 GB file
32K 427 GB · 128K 436 GB
Q5_K_S
Beam 1.0 501B A23B
345 GB355.1 GB file
32K 351 GB · 128K 368 GB
GLM 5.2
522 GB527.3 GB file
32K 521 GB · 128K 530 GB
BF16
Beam 1.0 501B A23B
1003 GB1002.7 GB file
32K 1009 GB · 128K 1026 GB
GLM 5.2
1511 GB1508.0 GB file
32K 1510 GB · 128K 1519 GB
Beam 1.0 501B A23B
Ollama · Q4_K_M
ollama run hf.co/bartowski/Beam-1.0-501B-A23B-GGUF:Q4_K_MGLM 5.2
Ollama · Q4_K_M
ollama run hf.co/unsloth/GLM-5.2-GGUF:Q4_K_M| Spec | Beam 1.0 501B A23B | GLM 5.2 |
|---|---|---|
| Publisher | reflection | Z.ai |
| Parameters | 501.3B | 753.3B |
| Active parameters | 23B per token | 42B per token |
| Architecture | BeamForCausalLM | GlmMoeDsaForCausalLM |
| Context | 1M tokens | 1M tokens |
| Modalities | Chat, Reasoning, Code, Function-calling | Chat, Function-calling |
| License | Apache 2.0 | MIT |
| Released | Oct 2026 | Jun 2026 |
Frequently Asked Questions
Which is better, Beam 1.0 501B A23B or GLM 5.2?
GLM 5.2 leads llmrun Score 122.9 to 119.5 (ranges overlap); neither fits 24 GB. Choose Beam 1.0 501B A23B for lighter on memory and also does reasoning, code; choose GLM 5.2 for higher llmrun Score and stronger at agents.
Which needs less VRAM, Beam 1.0 501B A23B or GLM 5.2?
At Q4_K_M, Beam 1.0 501B A23B requires 301.4 GB while GLM 5.2 needs 456.2 GB, making Beam 1.0 501B A23B the easier model to run on your local setup.
Can I run Beam 1.0 501B A23B or GLM 5.2 on a 16 GB GPU?
Beam 1.0 501B A23B does not fit (smallest quant needs 141.1 GB); GLM 5.2 does not fit (smallest quant needs 208.9 GB).
Which has a longer context window, Beam 1.0 501B A23B or GLM 5.2?
Beam 1.0 501B A23B supports 1,048,576 tokens and GLM 5.2 supports 1,048,576 tokens.