SIRumors
← Best frontier models
#23 frontier model

GLM 5.3

Z.ai · open weights
Frontier rank
#23
of 25 tracked
Frontier score
-1
100 = #1 on every board
Best placing
#16
Hard problems
Released
—
Not in the release tracker yet
GLM 5.3 against the field

Benchmarks

Frontier scoreAverage placing, 100 = #1 everywhere
  1. #1Claude Opus 5.597
  2. #2GPT-6 Astra95
  3. #3Claude Fable 5.184
  4. #4Gemini 4 Argon83
  5. #5GPT-6.1 Sol82
  6. #6Claude Opus 575
  7. #7GPT-6 Sol71
  8. #8Claude Fable 550
  9. #23GLM 5.3-1
OverallArena Elo
  1. #1Gemini 4 Argon1534
  2. #2Claude Opus 5.51515
  3. #3Claude Fable 5.11510
  4. #4Claude Opus 51507
  5. #5Claude Opus 4.61503
  6. #23GLM 5.31471
Hard problemsArena Elo
  1. #1Gemini 4 Argon1548
  2. #2Claude Opus 5.51535
  3. #3Claude Opus 4.61527
  4. #4Claude Opus 51526
  5. #5Claude Fable 5.11526
  6. #16GLM 5.31489
CodingArena Elo
  1. #1Claude Opus 5.51813
  2. #2GPT-6 Astra1786
  3. #3Claude Sonnet 5.51774
  4. #4GPT-6.1 Sol1755
  5. #5Claude Fable 5.11744
  6. #19GLM 5.31622
AgentsAgent score
  1. #1Claude Opus 5.514.3
  2. #2GPT-6 Astra13.1
  3. #3Claude Fable 5.112.7
  4. #4Claude Sonnet 5.512
  5. #5GPT-6.1 Sol11.7
  6. #23GLM 5.32.1
Arena leaderboards

Where it places

BoardPlacingScoreVotes
Overall#231471 Elo20,049
Hard problems#161489 Elo12,729
Coding#191622 Elo8,382
Agents#232.185,806

From blind head-to-head votes on Arena, published Oct 9, 2026. Variants (high, max) count once at their best.

News

Latest on GLM 5.3

↑ #22 Gemini 3.7 Flash#24 Claude Sonnet 5 ↓