vs
1:11
overall capability wins (0 ties across 12 scored). GLM-5.3 Flash leads overall.
- Reasoning1:4
- Coding0:3
- Arena0:1
- Tool use0:1
- Pricing0:3
- Speed2:0
- Specs0:2
Verdict
GLM-5.3 Flash leads coding (widest gap: +172.4 on WebDev Elo); GLM-5.3 Flash ranks higher on LMArena (+34 Elo); GLM-5.3 Flash offers ~7.8x larger context (1M vs 128K).
- GLM-5.3 Flash leads coding (widest gap: +172.4 on WebDev Elo)
- GLM-5.3 Flash ranks higher on LMArena (+34 Elo)
- GLM-5.3 Flash offers ~7.8x larger context (1M vs 128K)
- GLM-5.3 Flash allows ~7.8x more max output (128K vs 16K)
- GLM-5.3 Flash is ~1.5x cheaper on a blended token basis than GLM-4.7
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- GLM-4.7
- $0.40
- GLM-5.3 Flash
- $0.28
GLM-5.3 Flash is about 1.5x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Zhipu AI
Zhipu AI
Family
GLM
GLM
License
MIT
MIT
Open weights
Yes
Yes
Release
Dec 1, 2025
Aug 26, 2026
Knowledge cutoff
-
-
API / provider
Zhipu / Z.ai
Zhipu / Z.ai
Modalities
text → text
text, image, video → text
Specs
Context window
128K
1M
Max output
16K
128K
Parameters
—
320B (18B act.)
Pricing
Input $/1M
$0.20
$0.15
Output $/1M
$0.80
$0.50
Blended $/1M (3∶1)
$0.35
$0.24
Speed
tok/s
90
49.8
TTFT (s)
0.3
1.51
Reasoning
MMLU-Pro
—
—
GPQA Diamond
85.9
91.2
Humanity's Last Exam
27.4
39.9
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
29.3
27.5
AA-LCR v1.1
71.0
80.0
CritPt
1.7
15.4
MMMU-Pro
—
—
IFBench
67.9
—
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
—
Terminal-Bench 2.1
45.3
84.3
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
45.1
51.6
CursorBench
—
—
SWE-Rebench
45.5
—
NL2Repo-Bench
—
—
DeepSWE
—
63.0
WebDev Arena
1,435
1,607
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
1,442
1,475
| Metric | GLM-4.7 | GLM-5.3 Flash | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Zhipu AI | Zhipu AI | — |
| Family | GLM | GLM | — |
| License | MIT | MIT | — |
| Open weights | Yes | Yes | — |
| Release | Dec 1, 2025 | Aug 26, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Zhipu / Z.ai | Zhipu / Z.ai | — |
| Modalities | text → text | text, image, video → text | — |
| Specs | |||
| Context window | 128K | 1M | B +872K |
| Max output | 16K | 128K | B +112K |
| Parameters | — | 320B (18B act.) | — |
| Pricing | |||
| Input $/1M | $0.20 | $0.15 | B +$0.05 |
| Output $/1M | $0.80 | $0.50 | B +$0.30 |
| Blended $/1M (3∶1) | $0.35 | $0.24 | B +$0.11 |
| Speed | |||
| tok/s | 90 | 49.8 | A +40 |
| TTFT (s) | 0.3 | 1.51 | A +1.21 |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 85.9 | 91.2 | B +5.3 pts |
| Humanity's Last Exam | 27.4 | 39.9 | B +12.5 pts |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 29.3 | 27.5 | A +1.8 pts |
| AA-LCR v1.1 | 71.0 | 80.0 | B +9.0 pts |
| CritPt | 1.7 | 15.4 | B +13.7 pts |
| MMMU-Pro | — | — | — |
| IFBench | 67.9 | — | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | 45.3 | 84.3 | B +39.0 pts |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 45.1 | 51.6 | B +6.5 pts |
| CursorBench | — | — | — |
| SWE-Rebench | 45.5 | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | 63.0 | — |
| WebDev Arena | 1,435 | 1,607 | B +172 Elo |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | 1,442 | 1,475 | B +34 Elo |
GLM-4.7
- Context
- 128K / 16K out
- Parameters
- —
- Price
- $0.20 / $0.80
- Speed
- 90 tok/s · 0.3s TTFT
- Modalities
- text → text
- License
- MIT
Prior GLM-4.x open checkpoint still useful for cost-sensitive bilingual apps.
GLM-5.3 Flash
- Context
- 1M / 128K out
- Parameters
- 320B (18B act.)
- Price
- $0.15 / $0.50
- Speed
- 49.8 tok/s · 1.51s TTFT
- Modalities
- text, image, video → text
- License
- MIT
First natively multimodal GLM-5 (text/image/video in), MIT open weights, 320B total / 18B active MoE with 1M context, released Aug 26, 2026. List pricing $0.15/$0.50 per 1M (cache-hit ~$0.03), halved by a launch promo through Sept 9, 2026. Benchmark figures are Z.ai's self-reported launch numbers (DeepSWE 63.4, AutomationBench 48.8, Terminal-Bench 2.1 84.3); Artificial Analysis places it at 57 on its Intelligence Index.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.