Skip to content
Model Pareto

GLM-5.3

Z.ai (Zhipu)Open weightsz.ai ↗
Compare with nearest rivalsCompare with…
Input
$1.40/M tok
Output
$4.40/M tok
Blended
$2.15/M tok · 3:1
Speed
57tok/s
Context
1Mtokens

Open weights (744B-A40B).

Rankings

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
16× H200 141GB
FP8 weights
Throughput
~29,333 tok/s
many requests batched
Price
$0.28–2.33 /M tok
busy → light use
On your own machine
Mac Studio M5 Ultra 512GB (int4, ~42 tok/s single-stream)
single consumer GPU or Mac
Assumptions (5)
  • FP8 weights (744B params, 40B active per token) + 40% KV-cache headroom ≈ 1042 GB
  • 16× H200 141GB at $3.59–$7.91/GPU-hour on-demand (2026-09-24)
  • ~29,333 output tok/s aggregate at batch 1241 (bandwidth-bound); MoE compute scales with active params
  • Blended 3:1 input:output; prefill ~158,963 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
~#2 of 14 (estimated)~89.7%
  1. #1Claude Opus 596.0%
  2. ~#2GLM-5.3Estimated89.7%
  3. #2Claude Sonnet 585.2%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Estimated
~#5 of 19 (estimated)~71.3%
  1. #1Claude Opus 5.589.9%
  2. #2Claude Sonnet 5.581.3%
  3. #4Claude Opus 579.2%
  4. ~#5GLM-5.3Estimated71.3%
  5. #5Intern-S2-397B68.5%
No published score — estimated from related benchmarks
LiveCodeBenchEstimated
~#2 of 20 (estimated)~92.1%
  1. #1Qwen3.8-Omni-Flash92.6%
  2. ~#2GLM-5.3Estimated92.1%
  3. #2Qwen3.8 Flash-Next91.9%
No published score — estimated from related benchmarks

Writing & chat

IFBenchEstimated
~#3 of 126 (estimated)~81.7%
  1. #1MiniMax-M382.9%
  2. #2Qwen3.8 2.4T A95B82.8%
  3. ~#3GLM-5.3Estimated81.7%
  4. #3Qwen3.8-Omni-Flash81.5%
No published score — estimated from related benchmarks
MMMLUEstimated
~#2 of 5 (estimated)~91.3%
  1. #1Gemini 3.1 Pro92.6%
  2. ~#2GLM-5.3Estimated91.3%
  3. #2Gemma 4 31B88.4%
No published score — estimated from related benchmarks
LMArena Text, Non-English (Elo)Verified
#10 of 531466 Elo
  1. #1Claude Opus 5.51501
  2. #2Claude Fable 5.1Few votes1496
  3. #9Qwen3.8 Max (0902)1467
  4. #10GLM-5.31466
  5. #11GLM-5.3 Flash1463
lmarena.ai · 2026-09-24
LMArena Text (Elo)Verified
#10 of 531480 Elo
  1. #1Claude Opus 5.51509
  2. #2Claude Fable 5.11501
  3. #9Qwen3.8 Max (0902)1481
  4. #10GLM-5.31480
  5. #10GPT-6 AstraFew votes1480
lmarena.ai · 2026-09-25

Hard reasoning

AIME (latest)Estimated
~#2 of 25 (estimated)~96.7%
  1. #1Inkling97.1%
  2. ~#2GLM-5.3Estimated96.7%
  3. #2Muse Glimmer94.7%
No published score — estimated from related benchmarks

Agents

τ²-bench (Telecom)Estimated
~#2 of 116 (estimated)~98.3%
  1. #1Step 3.7 Flash98.5%
  2. ~#2GLM-5.3Estimated98.3%
  3. #2Gemini 3.1 Pro95.6%
No published score — estimated from related benchmarks
OSWorld-VerifiedEstimated
~#3 of 12 (estimated)~84.0%
  1. #1Kimi K384.8%
  2. #2Qwen3.8 27B84.3%
  3. ~#3GLM-5.3Estimated84.0%
  4. #3Nex-N2.5-Pro82.2%
No published score — estimated from related benchmarks
OSWorld 2.0Estimated
~#2 of 5 (estimated)~58.4%
  1. #1GLM-5.3 Flash59.1%
  2. ~#2GLM-5.3Estimated58.4%
  3. #2Kimi K358.3%
No published score — estimated from related benchmarks
BrowseCompEstimated
~#4 of 18 (estimated)~91.1%
  1. #1Atria Dawn Preview92.5%
  2. #2GPT-6 Astra91.5%
  3. #3Kimi K391.2%
  4. ~#4GLM-5.3Estimated91.1%
  5. #4Claude Opus 590.8%
No published score — estimated from related benchmarks