HolyGrAIL

All tasks / Models

GLM-5.3

Z.ai's open-weights GLM model line for reasoning, coding and agentic tasks.

  • Z.ai
  • Open weights
  • Released 14 Aug 2026

Official page

Where it ranks

  • A
    Chat & assistants
    Models · Local · 1,479 on LMArena Text Arena (overall, style control)
  • A
    Writing
    Models · Local · 1,453 on LMArena Text Arena: Creative Writing (style control)
  • C
    Coding
    Models · Local · 1,622 on LMArena WebDev / Code Arena (overall)
  • C
    Vector graphics & logos
    Models · Local · 1,293 on Design Arena SVG
  • B
    Workflow automation
    Models · Local · 62.2% on AutomationBench-AA (Artificial Analysis)
  • A
    Slides & presentations
    Models · Local · 1,255 on Design Arena Python-PPTX Slides

Facts

Vendor
Z.ai
Family
GLM
Released
14 Aug 2026
Weights
Open: download and run it yourself
License
GLM-5.3 License (custom; HF tag 'glm-5.3'; LMArena lists MIT)
Parameters
744B (40B active)
GPU memory to run locally
~425 GB (744B total / 40B active as reported in press coverage (HF safetensors total 753B); 4-bit ≈ 425 GB, estimate.)
API price
$1.4 in / $4.4 out per 1M tokens
Access
Free (open weights)
Last verified
18 Oct 2026

Price from openrouter.ai, 18 Oct 2026.

Benchmark results

BenchmarkResultEntry on the leaderboardAs ofSource
Artificial Analysis Intelligence Index45GLM-5.3 (Max)
Raw value from the page's embedded model data; the site displays it rounded to an integer.
18 Oct 2026artificialanalysis.ai
AutomationBench-AA (Artificial Analysis)62.2%GLM-5.3 (Max)
AutomationBench-AA Score (share of each task's objectives completed, a task with any guardrail violation scoring zero); raw value from the page's embedded model data (displayed rounded); rank 16 of 32; Tasks Completed 29.38%, Objectives Completed 87.17%; no last-updated date shown, so as_of = retrieval date.
19 Oct 2026artificialanalysis.ai
Design Arena Python-PPTX Slides1,255GLM-5.3
248 battles. Read from the site's own API (page renders client-side).
18 Oct 2026designarena.ai
Design Arena SVG1,293GLM-5.3
Design Arena SVG, rank 15, 1983 battles; board id 'cortado'; rows read from POST https://www.designarena.ai/api/leaderboard (arenaType 'models', category 'svg'); names from https://www.designarena.ai/api/registry
19 Oct 2026designarena.ai
EQ-Bench Creative Writing v32,075GLM-5.3
Elo score from the leaderboard data embedded in https://eqbench.com/creative_writing.js (file last modified 2026-09-24); rank 6 of 140; rubric score 17.04.
24 Sep 2026eqbench.com
LMArena Text Arena (overall, style control)1,479glm-5.3-max
From LMArena's official leaderboard-dataset (text_style_control, split latest, category overall, published 2026-10-15), which mirrors https://arena.ai/leaderboard/text (site blocks automated fetches).
15 Oct 2026huggingface.co
LMArena Text Arena: Creative Writing (style control)1,453glm-5.3-max
From LMArena's official leaderboard-dataset (text_style_control, split latest, category creative_writing, published 2026-10-15), which mirrors https://arena.ai/leaderboard/text/creative-writing (site blocks automated fetches). Rank 35 of 414; 5066 votes; 95% CI 1443.9-1462.5.
15 Oct 2026huggingface.co
LMArena WebDev / Code Arena (overall)1,622GLM-5.3 (Max)
From LMArena's official leaderboard-dataset (webdev = Code Arena, split latest, category overall, published 2026-10-18), which mirrors https://arena.ai/leaderboard/webdev. Rank 23; 9729 votes; 95% CI 1614.0-1629.2.
18 Oct 2026huggingface.co
SWE-Bench Pro V2 (public set, full)95.6%GLM-5.3 (mini-swe-agent) max
SWE-Bench Pro V2 Full (642 tasks); row added 2026-10-07; 95% CI +/-1.3; harness in the variant name; read from the leaderboard page's embedded data.
19 Oct 2026labs.scale.com

Where you can use it

Earlier versions

  • GLM-5.2 · —
  • GLM-5.1 · —
  • GLM-5 · 11 Feb 2026

Compare

Alternatives to GLM-5.3

Sources