Sign Up

GLM 5.3

Z.ai
Announcement

Cost/1M tokens

input
$0.0699
cached
$0.0689
output
$1.40
Context
1Mtokens
Released
3mo ago
Size
753Bparameters

Benchmarks

DeepSWE is the primary coding benchmark where an exact model evaluation exists. Arena leaderboards and other sourced evaluations appear below; missing scores are not zero.

How it ranks

Against its predecessor, similarly priced models and the models ranked around it.

DeepSWE 1.1

Best pass@1 per model · Higher is better

#7 of 28

  1. 74.1%
    GPT-6 Astra
  2. 73.8%
    Gemini 3.8 Flash
  3. 73.6%
    Claude Opus 5
  4. 72.7%
    GPT-5.6 Sol
  5. 69.9%
    Claude Fable 5
  6. 69.6%
    GPT-5.6 Terra
  7. 69.0%
    GLM 5.3
  8. 68.5%
    Kimi K3
  9. 67.5%
    Grok 4.6
  10. 67.2%
    GPT-5.6 Luna

Arena Text

Best Arena rating per model · Higher is better

Arena #26

  1. 1483
    Kimi K3.1 ProNearby rank
  2. 1483
    Gemini 3.6 FlashNearby rank
  3. 1483
    GLM 5.3
  4. 1482
    GPT-6.1 SolNearby rank
  5. 1482
    Claude Opus 4.8Nearby rank
  6. 1475
    GLM 5.2Predecessor
  7. 1417
    DeepSeek V3.1Similar price
  8. 1415
    MiniMax M2.7Similar price
  9. 1346
    Qwen-PlusSimilar price
  10. 1303
    Mistral Small 3.1 24BSimilar price

Level with GLM 5.2 (predecessor, overlapping 95% CIs) · Ahead of 4 of 4 similarly priced models

Arena WebDev

Best Arena rating per model · Higher is better

Arena #29

  1. 1624
    MiMo-V2.6-ProNearby rank
  2. 1624
    Qwen3.8 Flash NextNearby rank
  3. 1622
    GLM 5.3
  4. 1621
    DeepSeek V4.1 FlashNearby rank
  5. 1618
    Grok 4.6Nearby rank
  6. 1608
    GLM 5.2Predecessor
  7. 1519
    GPT-5.6 LunaSimilar price
  8. 1437
    Beam 1.0 501B A23BSimilar price
  9. 1398
    MiniMax M2.7Similar price
  10. 1253
    Qwen3.5-35B-A3BSimilar price

+15 vs GLM 5.2 (predecessor) · Ahead of 4 of 4 similarly priced models

Arena Search

Best Arena rating per model · Higher is better

Not ranked on Arena Search yet

Agent Arena

Best IPS score per model · Higher is better

Arena #27

  1. +0.024
    MiMo-V2.6-ProSimilar price
  2. +0.017
    Muse EmberNearby rank
  3. +0.016
    GPT-6 LunaNearby rank
  4. +0.010
    GLM 5.3
  5. +0.006
    Kimi K3Nearby rank
  6. +0.005
    Claude Haiku 5.5Nearby rank
  7. +0.002
    GLM 5.2Predecessor
  8. -0.059
    Beam 1.0 501B A23BSimilar price
  9. -0.080
    Qwen3.7 PlusSimilar price
  10. -0.105
    MiniMax M3Similar price

Level with GLM 5.2 (predecessor, overlapping 95% CIs) · Ahead of 3 of 4 similarly priced models

FrontierMath-Tier-4-v2-Private

Best score per model · Higher is better

#35 of 72

  1. 61.0%
    GPT-5.6 LunaSimilar price
  2. 29.3%
    Claude Sonnet 5Nearby rank
  3. 29.3%
    GLM 5.2Predecessor
  4. 29.3%
    GLM 5.3
  5. 26.8%
    Claude Opus 4.6Nearby rank
  6. 26.8%
    DeepSeek V4 Pro 0813Nearby rank
  7. 26.8%
    DeepSeek V4.1 FlashSimilar price
  8. 17.1%
    Inkling SmallSimilar price
  9. 12.2%
    GPT-5.4 NanoSimilar price

Level with GLM 5.2 (predecessor) · Ahead of 3 of 4 similarly priced models

FrontierMath-Tiers-1-3-v2-Private

Best score per model · Higher is better

#30 of 87

  1. 82.1%
    GPT-5.6 LunaSimilar price
  2. 71.2%
    DeepSeek V4.1 Flash 1017Nearby rank
  3. 70.2%
    Claude Opus 4.7Nearby rank
  4. 68.8%
    GLM 5.3
  5. 68.4%
    Gemini 3.8 FlashNearby rank
  6. 67.4%
    GPT-5.2Nearby rank
  7. 67.4%
    DeepSeek V4.1 FlashSimilar price
  8. 59.2%
    GLM 5.2Predecessor
  9. 27.7%
    Gemini 3.1 Flash LiteSimilar price
  10. 22.5%
    Qwen3.6 FlashSimilar price

+9.6 pts vs GLM 5.2 (predecessor) · Ahead of 3 of 4 similarly priced models

GPQA diamond

Best score per model · Higher is better

#33 of 161

  1. 91.9%
    GLM 5.2Predecessor
  2. 91.6%
    GPT-5.6 LunaSimilar price
  3. 90.9%
    MiniMax M3Similar price
  4. 90.9%
    Qwen3.7 MaxNearby rank
  5. 90.9%
    GLM 5.3
  6. 90.8%
    Kimi K2.6Nearby rank
  7. 90.7%
    GPT-5.5Nearby rank
  8. 71.2%
    DeepSeek V3Similar price
  9. 48.1%
    Qwen-PlusSimilar price

-0.9 pts vs GLM 5.2 (predecessor) · Ahead of 2 of 4 similarly priced models

OTIS Mock AIME 2024-2025

Best score per model · Higher is better

#54 of 158

  1. 98.3%
    GPT-5.6 LunaSimilar price
  2. 91.1%
    Qwen3.6 Max PreviewNearby rank
  3. 91.1%
    Qwen3.6 27BNearby rank
  4. 91.1%
    GLM 5.3
  5. 90.0%
    Inkling SmallNearby rank
  6. 88.9%
    GPT-5.4 MiniNearby rank
  7. 86.4%
    GLM 5.2Predecessor
  8. 71.1%
    MiniMax M3Similar price
  9. 48.9%
    DeepSeek V3Similar price
  10. 17.8%
    Qwen-PlusSimilar price

+4.7 pts vs GLM 5.2 (predecessor) · Ahead of 3 of 4 similarly priced models

SimpleQA Verified

Best score per model · Higher is better

#50 of 87

  1. 42.9%
    DeepSeek V4.1 FlashSimilar price
  2. 41.4%
    GPT-6 LunaNearby rank
  3. 41.0%
    GPT-5.6 LunaSimilar price
  4. 41.0%
    GLM 5.3
  5. 40.4%
    Qwen3 235B A22B Thinking 2507Nearby rank
  6. 40.3%
    InklingNearby rank
  7. 34.2%
    GLM 5.2Predecessor
  8. 19.1%
    Inkling SmallSimilar price
  9. 15.9%
    Qwen3.6 FlashSimilar price

+6.8 pts vs GLM 5.2 (predecessor) · Ahead of 2 of 4 similarly priced models

DeepSWE 1.1

Primary coding-agent benchmark · 113 tasks · Data generated 2026-09-22

Official leaderboard
69.0%
pass@1 · highest published configuration
mini-swe-agent · max reasoning
pass@4
87.6%
Average cost / attempt
$3.99
Benchmark runs
4

pass@1 is the success rate across scored attempts; pass@4 is the share of tasks solved in at least one of four attempts. Costs are reported by DeepSWE for each configuration. These results are not directly comparable to evaluations using different harnesses. Official source JSON.

All published scores

Benchmark compilation: Epoch AI, Capabilities & benchmarking (CC BY attribution). Scores are converted to percentages and model names matched to the Vertical Tabs catalog. Individual evaluations are credited to their original sources. External results may carry separate licences and evaluation configurations; do not interpret scores from different harnesses as directly comparable.

Arena ratings: Arena (lmarena-ai/leaderboard-dataset), licensed CC BY 4.0. Ratings come from pairwise human votes and are only comparable within one leaderboard; the interval is Arena's 95% confidence interval. Model names are matched to the Vertical Tabs catalog.

To give you the best experience, we use cookies for personalization, insights, and performance. Learn more.