GLM 4.6 vs GLM 5.3

GLM 5.3 leads GLM 4.6 on coding, 70.6 to 8.8 on the Vibe Coding Index, and GLM 4.6 costs 2.0x less per output token ($2.20 vs $4.40 per million). Updated October 21, 2026; 50 benchmarks have a result for both.

Replace GLM 4.6 Replace GLM 5.3

GLM 4.6Z.ai

Coding: 8.8
Overall: 33.4
Output $/M: $2.20

GLM 5.3Z.ai

Coding: 70.6
Overall: 67.9
Output $/M: $4.40

vs boards won of 50 shared

GLM 4.6 or GLM 5.3: which to pick

Pick GLM 5.3 for coding: it scores 70.6 on the Vibe Coding Index against 8.8 for GLM 4.6.

Pick GLM 4.6 for output tokens 2.0x cheaper ($2.20 vs $4.40 per million) and work measured by LiveCodeBench (Vals AI), where it leads.

Model AGLM 4.6Z.ai8.8/1002 independent benchmarksNo cost per solved task Model BGLM 5.3Z.ai70.6/1006 independent benchmarks$0.36 per solved task

501950

Shared results: 50
GLM 4.6 only: 19
GLM 5.3 only: 50
Like-for-like categories: 8 of 9

Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.

Which one for your work

Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.

Coding
Vibe Coding Index, six public coding benchmarks
GLM 5.3
GLM 5.3 scores 70.6 against 8.8.
2 and 6 benchmarks behind the scores
Intelligence
AA Intelligence Index
GLM 5.3
GLM 5.3 leads, 44.8 against 18.5.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only GLM 5.3 has a AA Agentic Index result, 53.1.
Directional only
Reasoning
GPQA Diamond
GLM 5.3
GLM 5.3 leads, 88.1% against 74.5%.
Same board, both measured
Math
AIME
No call
Only GLM 4.6 has a AIME result, 92.7%.
Directional only
Knowledge
MMLU Pro
GLM 5.3
GLM 5.3 leads, 86.8% against 82.2%.
Same board, both measured
Web design
Design Arena: websites
GLM 5.3
GLM 5.3 leads, 1295 against 1174.
Same board, both measured
General
Vals Index
No call
Only GLM 5.3 has a Vals Index result, 59.7%.
Directional only
Long documents
Listed context window
GLM 5.3
GLM 5.3 takes 1.05M tokens against 205K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
GLM 4.6
GLM 4.6 $0.0017, GLM 5.3 $0.0036 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only GLM 5.3 has the number, $0.36.
Directional only

At a glance

Coding index: GLM 4.6 8.8 from 2 benchmarks, GLM 5.3 70.6 from 6 benchmarks.
Widest category: Coding, 29.0 points apart on average over 8 shared benchmarks.
Widest single gap: Vibe Code Bench (Vals AI): GLM 4.6 3.1% against GLM 5.3 78.1%, 75.0 points apart on a 0 to 100 scale.
Output price: GLM 4.6 $2.20 against GLM 5.3 $4.40 per million output tokens, 2.0x apart.
Context window: GLM 4.6 takes 205K tokens, GLM 5.3 1.05M.

Benchmark by benchmark

First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.

Boards won, where both have a result
Category	GLM 4.6	GLM 5.3
Coding	1	7
Intelligence	0	1
Agentic work	0	1
Reasoning	0	6
Math	0	2
Knowledge	0	7
Web design	0	7
General	0	18
All shared boards	1	49

CodingLike-for-likeSix of these boards feed the Vibe Coding Index17

Coding benchmarks for GLM 4.6 and GLM 5.3
Benchmark	GLM 4.6	GLM 5.3	Leader
SWE-bench Verified	68.2%	Not measured	Only GLM 4.6 measured
LiveBench coding	Not measured	82.8%	Only GLM 5.3 measured
LMArena WebDev	1341	1621	GLM 5.3
AA Coding Index	45.8	74.8	GLM 5.3
LiveCodeBench (Vals AI)	81.0%	80.5%	GLM 4.6
SWE-bench Verified (Vals AI)	Not measured	95.4%	Only GLM 5.3 measured
Terminal-Bench 4 (Vals AI)	Not measured	38.9%	Only GLM 5.3 measured
Vibe Code Bench (Vals AI)	3.1%	78.1%	GLM 5.3
Code migration (Vals AI)	Not measured	44.2%	Only GLM 5.3 measured
ProgramBench (Vals AI)	Not measured	1.5%	Only GLM 5.3 measured
IOI (Vals AI)	Not measured	68.4%	Only GLM 5.3 measured
Terminal-Bench 2.1 (Vals AI)	Not measured	71.5%	Only GLM 5.3 measured
Vibe Code Bench 1-100 (Vals AI)	Not measured	20.0%	Only GLM 5.3 measured
SkillsBench (Vals AI)	Not measured	47.5%	Only GLM 5.3 measured
CyberBench (Vals AI)	Not measured	72.1%	Only GLM 5.3 measured
LiveCodeBench (Artificial Analysis run)	69.5%	Not measured	Only GLM 4.6 measured
SciCode	Not measured	59.0%	Only GLM 5.3 measured
Terminal-Bench Hard (Artificial Analysis run)	28.8%	Not measured	Only GLM 4.6 measured
Terminal-Bench	24.5%	Not measured	Only GLM 4.6 measured
FrontierCode	Not measured	40.1%	Only GLM 5.3 measured
FrontierSWE	Not measured	30.2%	Only GLM 5.3 measured
WeirdML	Not measured	75.4%	Only GLM 5.3 measured
ALE-Bench	340.8	1317.4	GLM 5.3
LMArena Text: coding	1449	1497	GLM 5.3
LMArena Text: software and IT	1455	1493	GLM 5.3
LMArena WebDev: React	Not measured	1641	Only GLM 5.3 measured
LMArena WebDev: HTML	1350	1595	GLM 5.3
Neither model has a result on the other 12 boards here

IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01

Intelligence benchmarks for GLM 4.6 and GLM 5.3
Benchmark	GLM 4.6	GLM 5.3	Leader
AA Intelligence Index	18.5	44.8	GLM 5.3
Epoch Capabilities Index	Not measured	156.0	Only GLM 5.3 measured

Agentic workLike-for-likePublic boards, shown as published, kept out of the index01

Agentic work benchmarks for GLM 4.6 and GLM 5.3
Benchmark	GLM 4.6	GLM 5.3	Leader
AA Agentic Index	Not measured	53.1	Only GLM 5.3 measured
Design Arena: agents building full-stack apps	1029	Not measured	Only GLM 4.6 measured
Design Arena: agents building mobile apps	1096	1194	GLM 5.3
Design Arena: agents building Android apps	1084	Not measured	Only GLM 4.6 measured
Design Arena: agents building Godot games	1180	Not measured	Only GLM 4.6 measured
Design Arena: HTML slides	Not measured	1189	Only GLM 5.3 measured
Design Arena: PowerPoint slides	Not measured	1254	Only GLM 5.3 measured
LiveBench research	Not measured	73.2%	Only GLM 5.3 measured
LiveBench browser	Not measured	0.0%	Only GLM 5.3 measured
LiveBench long agentic	Not measured	35.9%	Only GLM 5.3 measured
SREBench (Vals AI)	Not measured	9.2%	Only GLM 5.3 measured
Finance Agent (Vals AI)	Not measured	55.8%	Only GLM 5.3 measured
Tax Agent Bench (Vals AI)	Not measured	39.7%	Only GLM 5.3 measured
Legal Research Bench (Vals AI)	Not measured	49.0%	Only GLM 5.3 measured
Harvey's Legal Agent Benchmark (Vals AI)	Not measured	14.6%	Only GLM 5.3 measured
Excel Modeling Benchmark (Vals AI)	Not measured	56.3%	Only GLM 5.3 measured
Terminal-Bench-Science (Vals AI)	Not measured	5.7%	Only GLM 5.3 measured
τ²-Bench Telecom	76.9%	Not measured	Only GLM 4.6 measured
APEX-Agents	Not measured	56.6%	Only GLM 5.3 measured
Vending-Bench 2	Not measured	$8,164	Only GLM 5.3 measured
Berkeley Function Calling Leaderboard (BFCL)	72.4%	Not measured	Only GLM 4.6 measured
BrowseComp-Plus	48.2%	Not measured	Only GLM 4.6 measured
Neither model has a result on the other 19 boards here

ReasoningLike-for-likePublic boards, shown as published, kept out of the index06

Reasoning benchmarks for GLM 4.6 and GLM 5.3
Benchmark	GLM 4.6	GLM 5.3	Leader
GPQA Diamond	74.5%	88.1%	GLM 5.3
LiveBench reasoning	Not measured	85.8%	Only GLM 5.3 measured
MysteryMechanism (Vals AI)	Not measured	23.0%	Only GLM 5.3 measured
GPQA Diamond (Artificial Analysis run)	78.0%	91.7%	GLM 5.3
Humanity's Last Exam (Artificial Analysis run)	14.5%	42.3%	GLM 5.3
GPQA Diamond (Epoch AI run)	Not measured	90.9%	Only GLM 5.3 measured
Chess puzzles	Not measured	37.0%	Only GLM 5.3 measured
Mystery game puzzles	Not measured	33.0%	Only GLM 5.3 measured
LMArena Text: hard prompts	1440	1490	GLM 5.3
LMArena Text: expert	1432	1514	GLM 5.3
LMArena Text: hard prompts in English	1445	1496	GLM 5.3
Kagi LLM Benchmark	45.7%	Not measured	Only GLM 4.6 measured
Neither model has a result on the other 10 boards here

MathLike-for-likePublic boards, shown as published, kept out of the index02

Math benchmarks for GLM 4.6 and GLM 5.3
Benchmark	GLM 4.6	GLM 5.3	Leader
AIME	92.7%	Not measured	Only GLM 4.6 measured
MGSM (Vals AI)	89.8%	Not measured	Only GLM 4.6 measured
AA Math Index	86.0	Not measured	Only GLM 4.6 measured
AIME 2025 (Artificial Analysis run)	86.0%	Not measured	Only GLM 4.6 measured
OTIS Mock AIME	Not measured	91.1%	Only GLM 5.3 measured
FrontierMath, tiers 1 to 3	Not measured	68.8%	Only GLM 5.3 measured
FrontierMath, tier 4	Not measured	29.3%	Only GLM 5.3 measured
ProofBench	Not measured	49.0%	Only GLM 5.3 measured
LMArena Text: math	1432	1495	GLM 5.3
LMArena Text: mathematical work	1447	1497	GLM 5.3
Neither model has a result on the other 3 boards here

KnowledgeLike-for-likePublic boards, shown as published, kept out of the index07

Knowledge benchmarks for GLM 4.6 and GLM 5.3
Benchmark	GLM 4.6	GLM 5.3	Leader
MMLU Pro	82.2%	86.8%	GLM 5.3
LegalBench (Vals AI)	79.6%	84.8%	GLM 5.3
CorpFin (Vals AI)	56.8%	Not measured	Only GLM 4.6 measured
TaxEval (Vals AI)	66.2%	72.4%	GLM 5.3
MedQA (Vals AI)	92.2%	Not measured	Only GLM 4.6 measured
MedCode (Vals AI)	Not measured	42.9%	Only GLM 5.3 measured
MMLU Pro (Artificial Analysis run)	82.9%	Not measured	Only GLM 4.6 measured
SimpleQA Verified	Not measured	41.0%	Only GLM 5.3 measured
LMArena Text: science	1456	1498	GLM 5.3
LMArena Text: medicine and healthcare	1455	1481	GLM 5.3
LMArena Text: legal and government	1448	1473	GLM 5.3
LMArena Text: business and finance	1416	1459	GLM 5.3
Neither model has a result on the other 6 boards here

Web designLike-for-likePublic boards, shown as published, kept out of the index07

Web design benchmarks for GLM 4.6 and GLM 5.3
Benchmark	GLM 4.6	GLM 5.3	Leader
Design Arena: websites	1174	1295	GLM 5.3
Design Arena: UI components	1148	1314	GLM 5.3
Design Arena: game dev	1151	1311	GLM 5.3
Design Arena: data viz	1161	1259	GLM 5.3
Design Arena: 3D	1129	1332	GLM 5.3
Design Arena: all code categories	1166	1304	GLM 5.3
Design Arena: SVG	1109	1293	GLM 5.3
Design Arena: ASCII art	Not measured	1213	Only GLM 5.3 measured
LiveBench design	Not measured	77.3%	Only GLM 5.3 measured

GeneralLike-for-likePublic boards, shown as published, kept out of the index018

General benchmarks for GLM 4.6 and GLM 5.3
Benchmark	GLM 4.6	GLM 5.3	Leader
Vals Index	Not measured	59.7%	Only GLM 5.3 measured
LiveBench data analysis	Not measured	70.2%	Only GLM 5.3 measured
Public Benefits Bench (Vals AI)	Not measured	68.5%	Only GLM 5.3 measured
MedScribe (Vals AI)	Not measured	88.8%	Only GLM 5.3 measured
IFBench	43.4%	Not measured	Only GLM 4.6 measured
Long context reasoning (AA-LCR)	54.0%	79.7%	GLM 5.3
DTBench	Not measured	87.7%	Only GLM 5.3 measured
LMCA	Not measured	55.5%	Only GLM 5.3 measured
LMArena Text	1440	1474	GLM 5.3
LMArena Text: instruction following	1410	1479	GLM 5.3
LMArena Text: creative writing	1412	1458	GLM 5.3
LMArena Text: English	1449	1482	GLM 5.3
LMArena Text: non-English	1426	1459	GLM 5.3
LMArena Text: Chinese	1499	1532	GLM 5.3
LMArena Text: German	1447	1488	GLM 5.3
LMArena Text: Japanese	1393	1460	GLM 5.3
LMArena Text: Korean	1400	1460	GLM 5.3
LMArena Text: French	1459	1503	GLM 5.3
LMArena Text: Spanish	1436	1462	GLM 5.3
LMArena Text: Russian	1418	1469	GLM 5.3
LMArena Text: Polish	1447	1466	GLM 5.3
LMArena Text: multi-turn	1427	1472	GLM 5.3
LMArena Text: longer queries	1422	1483	GLM 5.3
LMArena Text: writing and language	1415	1469	GLM 5.3
LMArena Text: entertainment and media	1409	1447	GLM 5.3
Neither model has a result on the other 21 boards here

Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared

Image generation benchmarks for GLM 4.6 and GLM 5.3
Benchmark	GLM 4.6	GLM 5.3	Leader
Neither model has a result on 5 boards here

What each workload costs

Three fixed token mixes at each model's listed price, so per-token rates become a bill.

Chat turn

1K input + 500 output tokens

GLM 4.6$0.0017Fits in one request

GLM 5.3$0.0036Fits in one request

GLM 4.6 costs less

Repository review

50K input + 3K output tokens

GLM 4.6$0.037Fits in one request

GLM 5.3$0.083Fits in one request

GLM 4.6 costs less

Agent loop

200K input + 10K output tokens

GLM 4.6$0.142Exceeds the context window

GLM 5.3$0.324Fits in one request

GLM 4.6 costs less

Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.

Specification differences

The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.

Specifications of GLM 4.6 and GLM 5.3
Specification	Z.ai GLM 4.6	Z.ai GLM 5.3
Maker	Z.ai	Z.ai
Weights	Open weights	Open weights
Released	September 30, 2025	August 18, 2026
Context window	205K	1.05M
Input $/Mtok	$0.60	$1.40
Output $/Mtok	$2.20	$4.40
Coding index	8.8	70.6
Independent measurements behind it	2	6
Cost per solved task	Not computable	$0.36
Source	OpenRouter listing	OpenRouter listing

Questions

Is GLM 4.6 better than GLM 5.3 for coding?

Pick GLM 5.3 for coding: it scores 70.6 on the Vibe Coding Index against 8.8 for GLM 4.6. Pick GLM 4.6 for output tokens 2.0x cheaper ($2.20 vs $4.40 per million) and work measured by LiveCodeBench (Vals AI), where it leads.

Which is cheaper, GLM 4.6 or GLM 5.3?

GLM 4.6 costs $0.60 per million input tokens and $2.20 per million output tokens; GLM 5.3 costs $1.40 and $4.40. Cost per solved task is known for GLM 5.3 only.

Which has the bigger context window, GLM 4.6 or GLM 5.3?

GLM 5.3, with 1.05M tokens against 205K.

How many benchmarks is the GLM 4.6 vs GLM 5.3 comparison based on?

50 benchmarks have a result for both models as of October 21, 2026. GLM 4.6's coding index rests on 2 independent benchmarks and GLM 5.3's on 6. Where only one model has a result, the page says so instead of naming a winner.

Compare next

Embed this comparison on your site

Free under CC BY 4.0, while the source link stays visible. Preview the frame · Terms and other embeds

<iframe src="https://llmbenchmark.ai/embed/compare/glm-4-6-vs-glm-5-3" title="GLM 4.6 vs GLM 5.3, LLMBenchmark.ai" width="100%" height="360" style="border:0;max-width:100%" loading="lazy"></iframe>
<p style="font:13px/1.4 sans-serif;margin:6px 0 0">Source: <a href="https://llmbenchmark.ai/compare/glm-4-6-vs-glm-5-3">LLMBenchmark.ai, GLM 4.6 vs GLM 5.3</a></p>

GLM 4.6 or GLM 5.3: which to pick

Which one for your work

Coding

Intelligence

Agentic work

Reasoning

Math

Knowledge

Web design

General

Long documents

Everyday cost

Cost per solved task

At a glance

Benchmark by benchmark

What each workload costs

Chat turn

Repository review

Agent loop

Specification differences

Questions

Compare next