GLM-5.3 Flash
- Input
- $0.15/M tok
- Output
- $0.50/M tok
- Blended
- $0.237/M tok · 3:1
- Speed
- 64tok/s
- Context
- 1Mtokens
- Cost per task
- ~$0.21/taskest.
Open weights (320B-A18B, MIT), natively multimodal.
Rankings
Coding
Mixed#32 / 137rank by quality66.1±6.6quality / 1003/7 benchmarks measuredWriting & chat
Verified#22 / 162rank by quality74.2±6.0quality / 1002/5 benchmarks measuredVision
Verified#42 / 98rank by quality60.0±7.4quality / 1002/6 benchmarks measuredHard reasoning
Verified#32 / 218rank by quality73.3±3.7quality / 1003/4 benchmarks measuredAgents
Mixed#11 / 187rank by quality82.9±5.2quality / 1004/7 benchmarks measuredPrice frontier
Self-hostingEstimated
How we estimate →What it would cost to run these open weights yourself on rented GPUs, compared with the API price above.
- Hardware
- 4× B200 180GB
- FP8 weights
- Throughput
- ~16,470 tok/s
- many requests batched
- Price
- $0.16–1.45 /M tok
- busy → light use
- On your own machine
- Mac Studio M5 Ultra 512GB (int4, ~93 tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- FP8 weights (320B params, 18B active per token) + 40% KV-cache headroom ≈ 448 GB
- 4× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
- ~16,470 output tok/s aggregate at batch 471 (bandwidth-bound); MoE compute scales with active params
- Blended 3:1 input:output; prefill ~236,250 tok/s
- Low = 75% utilization at the low GPU price; high = 20% at the high price
Benchmark scores
~ italic, dashed = no published score yet, estimated from related benchmarks.
Coding
SWE-bench VerifiedEstimated
~#3 of 21 (estimated)~84.2%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Estimated
~#7 of 24 (estimated)~67.3%
- #1Claude Fable 5.293.0%
- #2Claude Opus 5.589.9%
- #6Intern-S2-397B68.5%
- ~#7GLM-5.3 FlashEstimated67.3%
- #7Beam 1.0 501B A23B66.1%
No published score — estimated from related benchmarks
DeepSWE v1.1Lab-reported
#25 of 4563.4%
- #1Gemini 4 Argon78.8%
- #2DeepSeek V4.1 Flash 101777.0%
- #24GPT-6 Luna63.7%
- #25GLM-5.3 Flash63.4%
- #26Claude Opus 562.5%
z.ai · 2026-08-26
Terminal-Bench 4.0Verified
#22 of 10132.8%
- #1Kimi K3.1 Pro67.2%
- #2Claude Sonnet 5.563.6%
- #22Claude Haiku 5.532.8%
- #22GLM-5.3 Flash32.8%
- #24DeepSeek V4.1 Flash26.8%
artificialanalysis.ai · 2026-09-24
LiveCodeBenchEstimated
~#11 of 28 (estimated)~82.5%
- #1Qwen3.8-Omni-Flash92.6%
- #2Qwen3.8 Flash-Next91.9%
- #10Ternary Bonsai 27B82.8%
- ~#11GLM-5.3 FlashEstimated82.5%
- #11Gemma 4 31B80.0%
No published score — estimated from related benchmarks
SciCodeVerified
#41 of 11851.6%
- #1Claude Fable 5.267.7%
- #2Claude Opus 5.566.9%
- #40DeepSeek V4.1 Flash51.9%
- #41GLM-5.3 Flash51.6%
- #42MiMo-V2.6-Flash51.3%
artificialanalysis.ai · 2026-09-24
AA Coding Agent IndexEstimated
~#18 of 26 (estimated)~44.7
- #1Claude Opus 5.561.7
- #2Claude Opus 559.7
- #17Muse Spark 1.345.7
- ~#18GLM-5.3 FlashEstimated44.7
- #18DeepSeek V4 Pro (0813)43.1
No published score — estimated from related benchmarks
Writing & chat
IFBenchEstimated
~#31 of 140 (estimated)~71.5%
- #1MiniMax-M382.9%
- #2Qwen3.8 2.4T A95B82.8%
- #30Gemma 4 26B A4B72.4%
- ~#31GLM-5.3 FlashEstimated71.5%
- #31Nemotron 3 Super 120B A12B71.5%
No published score — estimated from related benchmarks
LMArena Text, Non-English (Elo)Verified
#12 of 561463 Elo
- #1Gemini 4 ArgonFew votes1512
- #2Claude Opus 5.51497
- #12GLM-5.31463
- #12GLM-5.3 Flash1463
- #12GPT-6 AstraFew votes1463
lmarena.ai · 2026-09-24
LMArena Text (Elo)Verified
#15 of 561475 Elo
- #1Gemini 4 Argon1525
- #2Claude Opus 5.51507
- #14GPT-6 AstraFew votes1477
- #15GLM-5.3 Flash1475
- #16DeepSeek V4.1 Flash1474
lmarena.ai · 2026-09-24
EQ-Bench Creative Writing v3 (Elo)Estimated
~#9 of 26 (estimated)~1919 Elo
- #1GPT-6 Astra2173
- #2Claude Fable 5.12162
- #8Grok 4.72007
- ~#9GLM-5.3 FlashEstimated1919
- #9Muse Spark 1.31906
No published score — estimated from related benchmarks
Vision
MMMU-ProEstimated
~#20 of 88 (estimated)~79.1%
- #1Claude Opus 5.587.7%
- #2GPT-6 Astra86.9%
- #19MiniMax-M3.1-Flash79.1%
- ~#20GLM-5.3 FlashEstimated79.1%
- #20Gemini 3.5 Flash-Lite79.0%
No published score — estimated from related benchmarks
CharXiv ReasoningEstimated
~#6 of 12 (estimated)~81.6%
- #1Gemini 3.8 Flash86.2%
- #2Kimi K384.8%
- #5MiniMax-M3.1-Flash82.5%
- ~#6GLM-5.3 FlashEstimated81.6%
- #6Nemotron 3.5 Super VL80.8%
No published score — estimated from related benchmarks
OmniDocBench v1.5Estimated
~#4 of 8 (estimated)~87.4%
- #1MiniMax-M391.6%
- #2Kimi K391.1%
- #2Qwen3.8 27B91.1%
- ~#4GLM-5.3 FlashEstimated87.4%
- #4Gemini 3.1 Pro85.3%
No published score — estimated from related benchmarks
GDP.pdf (AA)Verified
#24 of 4715.4%
- #1Muse Wonder Full32.6%
- #2GPT-6 Astra31.0%
- #23Qwen3.8 Flash-Next15.6%
- #24GLM-5.3 Flash15.4%
- #25Step 5 Preview14.8%
artificialanalysis.ai · 2026-09-24
ChartographyVerified
#19 of 2416.3%
- #1Claude Fable 5.273.4%
- #2GPT-6 Astra71.0%
- #18Claude Sonnet 516.6%
- #19GLM-5.3 Flash16.3%
- #20Grok 4.714.7%
surgehq.ai · 2026-09-24
Hard reasoning
GPQA DiamondVerified
#25 of 19291.2%
- #1GPT-6 Astra96.1%
- #2Gemini 3.8 Flash95.3%
- #24DeepSeek V4.1 Flash NVFP491.3%
- #25GLM-5.3 Flash91.2%
- #26Claude Sonnet 591.1%
artificialanalysis.ai · 2026-09-24
Humanity's Last ExamVerified
#34 of 19839.9%
- #1Claude Fable 5.274.3%
- #2Claude Opus 5.561.4%
- #33DeepSeek V4 Pro (0813)41.0%
- #34GLM-5.3 Flash39.9%
- #35Ling 3.1 Flash39.4%
artificialanalysis.ai · 2026-09-24
AIME (latest)Estimated
~#3 of 34 (estimated)~97.6%
No published score — estimated from related benchmarks
Agents
τ²-bench (Telecom)Estimated
~#6 of 120 (estimated)~94.4%
- #1Step 3.7 Flash98.5%
- #2DeepSeek-V4-Pro-0813-nvfp4-DSpark98.3%
- #5Qwen3.6 35B A3B95.3%
- ~#6GLM-5.3 FlashEstimated94.4%
- #6Mistral Medium 3.594.2%
No published score — estimated from related benchmarks
OSWorld-VerifiedEstimated
~#3 of 15 (estimated)~82.4%
No published score — estimated from related benchmarks
BrowseCompEstimated
~#6 of 22 (estimated)~89.5%
- #1Atria Dawn Preview92.5%
- #2GPT-6 Astra91.5%
- #5Nex-N2.5-Pro89.7%
- ~#6GLM-5.3 FlashEstimated89.5%
- #6Step 5 Preview88.7%
No published score — estimated from related benchmarks
AutomationBench-AAVerified
#12 of 8460.4%
- #1Claude Fable 5.276.4%
- #2Muse Wonder Full71.3%
- #11GPT-6 Sol61.6%
- #12GLM-5.3 Flash60.4%
- #13Gemini 3.8 Flash59.9%
artificialanalysis.ai · 2026-09-24