Enter opens the highlighted result. Esc closes.

Gemma 4 31B vs GLM 5.2

GLM 5.2 leads Gemma 4 31B on coding, 61.1 to 39.0 on the Vibe Coding Index, and Gemma 4 31B costs 11.0x less per output token ($0.40 vs $4.40 per million). Updated October 22, 2026; 48 benchmarks have a result for both.

Replace Gemma 4 31B Replace GLM 5.2
Gemma 4 31BGoogle
Coding
39.0
Overall
42.1
Output $/M
$0.40
2
GLM 5.2Z.ai
Coding
61.1
Overall
62.5
Output $/M
$4.40
46
vs boards won of 48 shared

Gemma 4 31B or GLM 5.2: which to pick

Pick GLM 5.2 for coding: it scores 61.1 on the Vibe Coding Index against 39.0 for Gemma 4 31B.

Pick Gemma 4 31B for output tokens 11.0x cheaper ($0.40 vs $4.40 per million) and work measured by Kagi LLM Benchmark or IFBench, where it leads.

Model AGemma 4 31BGoogle39.0/1002 independent benchmarksNo cost per solved taskModel BGLM 5.2Z.ai61.1/1004 independent benchmarks$0.86 per solved task
Shared results
48
Gemma 4 31B only
23
GLM 5.2 only
50
Like-for-like categories
7 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.

At a glance

Coding index
Gemma 4 31B 39.0 from 2 benchmarks, GLM 5.2 61.1 from 4 benchmarks.
Widest category
Agentic work, 33.9 points apart on average over 2 shared benchmarks.
Widest single gap
LMArena WebDev: React: Gemma 4 31B 1359 against GLM 5.2 1617, 40.1 points apart on a 0 to 100 scale.
Output price
Gemma 4 31B $0.40 against GLM 5.2 $4.40 per million output tokens, 11.0x apart.
Context window
Gemma 4 31B takes 262K tokens, GLM 5.2 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
CategoryGemma 4 31BBarsGLM 5.2Ties
Coding010
Intelligence02
Agentic work02
Reasoning16
Math03
Knowledge05
General118
All shared boards246
CodingLike-for-likeSix of these boards feed the Vibe Coding Index010
Coding benchmarks for Gemma 4 31B and GLM 5.2
BenchmarkGemma 4 31BGLM 5.2Leader
LiveBench coding49.0%Not measuredOnly Gemma 4 31B measured
LMArena WebDev13671604GLM 5.2
AA Coding Index43.468.8GLM 5.2
LiveCodeBench (Vals AI)Not measured69.5%Only GLM 5.2 measured
SWE-bench Verified (Vals AI)Not measured82.8%Only GLM 5.2 measured
Vibe Code Bench (Vals AI)Not measured64.0%Only GLM 5.2 measured
Code migration (Vals AI)Not measured37.9%Only GLM 5.2 measured
ProgramBench (Vals AI)Not measured0.5%Only GLM 5.2 measured
Terminal-Bench 2.1 (Vals AI)Not measured67.8%Only GLM 5.2 measured
SkillsBench (Vals AI)Not measured45.1%Only GLM 5.2 measured
SciCode45.5%51.2%GLM 5.2
Terminal-Bench Hard (Artificial Analysis run)36.4%50.8%GLM 5.2
FrontierCodeNot measured24.5%Only GLM 5.2 measured
WeirdML52.3%70.1%GLM 5.2
ALE-Bench925.51047.0GLM 5.2
LMArena Text: coding14591487GLM 5.2
LMArena Text: software and IT14631486GLM 5.2
LMArena WebDev: React13591617GLM 5.2
LMArena WebDev: HTML13801565GLM 5.2
SciCode8.9%Not measuredOnly Gemma 4 31B measured
Neither model has a result on the other 19 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for Gemma 4 31B and GLM 5.2
BenchmarkGemma 4 31BGLM 5.2Leader
AA Intelligence Index14.733.7GLM 5.2
Epoch Capabilities Index143.0151.9GLM 5.2
Agentic workLike-for-likePublic boards, shown as published, kept out of the index02
Agentic work benchmarks for Gemma 4 31B and GLM 5.2
BenchmarkGemma 4 31BGLM 5.2Leader
AA Agentic Index4.238.4GLM 5.2
Design Arena: agents building web appsNot measured1222Only GLM 5.2 measured
Design Arena: agents building full-stack appsNot measured1241Only GLM 5.2 measured
Design Arena: agents building mobile appsNot measured1159Only GLM 5.2 measured
Design Arena: agents building Android appsNot measured1178Only GLM 5.2 measured
Design Arena: agents building Godot gamesNot measured1142Only GLM 5.2 measured
Design Arena: agents building gamesNot measured1206Only GLM 5.2 measured
Design Arena: HTML slidesNot measured1125Only GLM 5.2 measured
Design Arena: PowerPoint slidesNot measured1170Only GLM 5.2 measured
LiveBench research44.4%Not measuredOnly Gemma 4 31B measured
LiveBench browser4.0%Not measuredOnly Gemma 4 31B measured
LiveBench long agentic13.3%Not measuredOnly Gemma 4 31B measured
SREBench (Vals AI)Not measured0.0%Only GLM 5.2 measured
Finance Agent (Vals AI)Not measured49.7%Only GLM 5.2 measured
Legal Research Bench (Vals AI)Not measured31.3%Only GLM 5.2 measured
Excel Modeling Benchmark (Vals AI)Not measured61.5%Only GLM 5.2 measured
τ²-Bench Telecom65.5%99.1%GLM 5.2
APEX-AgentsNot measured45.2%Only GLM 5.2 measured
PostTrainBenchNot measured31.7%Only GLM 5.2 measured
Vending-Bench 2Not measured$8,314Only GLM 5.2 measured
Neither model has a result on the other 21 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index16
Reasoning benchmarks for Gemma 4 31B and GLM 5.2
BenchmarkGemma 4 31BGLM 5.2Leader
GPQA DiamondNot measured85.6%Only GLM 5.2 measured
LiveBench reasoning68.3%Not measuredOnly Gemma 4 31B measured
GPQA Diamond (Artificial Analysis run)85.7%89.5%GLM 5.2
Humanity's Last Exam (Artificial Analysis run)23.6%41.1%GLM 5.2
GPQA Diamond (Epoch AI run)75.8%91.9%GLM 5.2
ARC-AGINot measured77.0%Only GLM 5.2 measured
ARC-AGI-2Not measured22.8%Only GLM 5.2 measured
SimpleBenchNot measured58.8%Only GLM 5.2 measured
Chess puzzlesNot measured22.0%Only GLM 5.2 measured
Mystery game puzzlesNot measured19.0%Only GLM 5.2 measured
EBR-benchNot measured9.5%Only GLM 5.2 measured
LMArena Text: hard prompts14481481GLM 5.2
LMArena Text: expert14641487GLM 5.2
LMArena Text: hard prompts in English14561481GLM 5.2
LMArena Vision: diagrams1283Not measuredOnly Gemma 4 31B measured
LMArena Vision: homework1310Not measuredOnly Gemma 4 31B measured
Kagi LLM Benchmark63.5%62.6%Gemma 4 31B
Neither model has a result on the other 5 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index03
Math benchmarks for Gemma 4 31B and GLM 5.2
BenchmarkGemma 4 31BGLM 5.2Leader
OTIS Mock AIME73.3%86.4%GLM 5.2
FrontierMath, tiers 1 to 3Not measured59.2%Only GLM 5.2 measured
FrontierMath, tier 4Not measured29.3%Only GLM 5.2 measured
ProofBenchNot measured35.0%Only GLM 5.2 measured
LMArena Text: math14631483GLM 5.2
LMArena Text: mathematical work14651489GLM 5.2
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index05
Knowledge benchmarks for Gemma 4 31B and GLM 5.2
BenchmarkGemma 4 31BGLM 5.2Leader
MMLU ProNot measured86.7%Only GLM 5.2 measured
LegalBench (Vals AI)Not measured84.1%Only GLM 5.2 measured
CaseLaw (Vals AI)52.6%Not measuredOnly Gemma 4 31B measured
CorpFin (Vals AI)Not measured66.1%Only GLM 5.2 measured
TaxEval (Vals AI)Not measured73.3%Only GLM 5.2 measured
MortgageTax (Vals AI)61.4%Not measuredOnly Gemma 4 31B measured
MedCode (Vals AI)Not measured40.8%Only GLM 5.2 measured
SimpleQA Verified10.4%34.2%GLM 5.2
LMArena Text: science14571489GLM 5.2
LMArena Text: medicine and healthcare14511475GLM 5.2
LMArena Text: legal and government14381486GLM 5.2
LMArena Text: business and finance14301451GLM 5.2
FACTS Benchmark Suite45.3%Not measuredOnly Gemma 4 31B measured
Neither model has a result on the other 5 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index9 boards, none shared
Web design benchmarks for Gemma 4 31B and GLM 5.2
BenchmarkGemma 4 31BGLM 5.2Leader
Design Arena: websitesNot measured1285Only GLM 5.2 measured
Design Arena: UI componentsNot measured1294Only GLM 5.2 measured
Design Arena: game devNot measured1268Only GLM 5.2 measured
Design Arena: data vizNot measured1285Only GLM 5.2 measured
Design Arena: 3DNot measured1287Only GLM 5.2 measured
Design Arena: all code categoriesNot measured1287Only GLM 5.2 measured
Design Arena: SVGNot measured1213Only GLM 5.2 measured
Design Arena: ASCII artNot measured1191Only GLM 5.2 measured
LiveBench design58.4%Not measuredOnly Gemma 4 31B measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index118
General benchmarks for Gemma 4 31B and GLM 5.2
BenchmarkGemma 4 31BGLM 5.2Leader
LiveBench data analysis53.3%Not measuredOnly Gemma 4 31B measured
SAGE (Vals AI)55.0%Not measuredOnly Gemma 4 31B measured
MedScribe (Vals AI)Not measured83.5%Only GLM 5.2 measured
IFBench75.6%73.3%Gemma 4 31B
Long context reasoning (AA-LCR)69.7%78.3%GLM 5.2
DTBench82.7%93.6%GLM 5.2
LMCA39.3%45.8%GLM 5.2
Surface Evolver Bench30.6%55.6%GLM 5.2
LMArena Text14431470GLM 5.2
LMArena Text: instruction following14331466GLM 5.2
LMArena Text: creative writing14171463GLM 5.2
LMArena Vision1276Not measuredOnly Gemma 4 31B measured
LMArena Text: English14511475GLM 5.2
LMArena Text: non-English14301459GLM 5.2
LMArena Text: Chinese14751522GLM 5.2
LMArena Text: GermanNot measured1470Only GLM 5.2 measured
LMArena Text: JapaneseNot measured1450Only GLM 5.2 measured
LMArena Text: KoreanNot measured1440Only GLM 5.2 measured
LMArena Text: French14361479GLM 5.2
LMArena Text: Spanish14421478GLM 5.2
LMArena Text: Russian14601468GLM 5.2
LMArena Text: Polish14331456GLM 5.2
LMArena Text: multi-turn14521469GLM 5.2
LMArena Text: longer queries14461480GLM 5.2
LMArena Text: writing and language14321463GLM 5.2
LMArena Text: entertainment and media14051450GLM 5.2
LMArena Vision: English1271Not measuredOnly Gemma 4 31B measured
LMArena Vision: Chinese1326Not measuredOnly Gemma 4 31B measured
LMArena Vision: reading text1282Not measuredOnly Gemma 4 31B measured
LMArena Vision: humor1280Not measuredOnly Gemma 4 31B measured
LMArena Vision: captioning1218Not measuredOnly Gemma 4 31B measured
LMArena Vision: recognizing things1226Not measuredOnly Gemma 4 31B measured
LMArena Vision: creative writing1285Not measuredOnly Gemma 4 31B measured
LMArena Document1425Not measuredOnly Gemma 4 31B measured
Neither model has a result on the other 12 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Gemma 4 31B and GLM 5.2
BenchmarkGemma 4 31BGLM 5.2Leader
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

Gemma 4 31B$0.0003Fits in one request
GLM 5.2$0.0036Fits in one request

Gemma 4 31B costs less

Repository review

50K input + 3K output tokens

Gemma 4 31B$0.0082Fits in one request
GLM 5.2$0.083Fits in one request

Gemma 4 31B costs less

Agent loop

200K input + 10K output tokens

Gemma 4 31B$0.032Fits in one request
GLM 5.2$0.324Fits in one request

Gemma 4 31B 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 Gemma 4 31B and GLM 5.2
SpecificationGoogle Gemma 4 31BZ.ai GLM 5.2
MakerGoogleZ.ai
WeightsOpen weightsOpen weights
ReleasedApril 2, 2026June 16, 2026
Context window262K1.05M
Input $/Mtok$0.14$1.40
Output $/Mtok$0.40$4.40
Coding index39.061.1
Independent measurements behind it24
Cost per solved taskNot computable$0.86
SourceOpenRouter listingOpenRouter listing

Questions

Is Gemma 4 31B better than GLM 5.2 for coding?

Pick GLM 5.2 for coding: it scores 61.1 on the Vibe Coding Index against 39.0 for Gemma 4 31B. Pick Gemma 4 31B for output tokens 11.0x cheaper ($0.40 vs $4.40 per million) and work measured by Kagi LLM Benchmark or IFBench, where it leads.

Which is cheaper, Gemma 4 31B or GLM 5.2?

Gemma 4 31B costs $0.14 per million input tokens and $0.40 per million output tokens; GLM 5.2 costs $1.40 and $4.40. Cost per solved task is known for GLM 5.2 only.

Which has the bigger context window, Gemma 4 31B or GLM 5.2?

GLM 5.2, with 1.05M tokens against 262K.

How many benchmarks is the Gemma 4 31B vs GLM 5.2 comparison based on?

48 benchmarks have a result for both models as of October 22, 2026. Gemma 4 31B's coding index rests on 2 independent benchmarks and GLM 5.2's on 4. 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/gemma-4-31b-it-vs-glm-5-2" title="Gemma 4 31B vs GLM 5.2, 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/gemma-4-31b-it-vs-glm-5-2">LLMBenchmark.ai, Gemma 4 31B vs GLM 5.2</a></p>