Enter opens the highlighted result. Esc closes.

GLM 5.1 vs Qwen2.5 72B Instruct

GLM 5.1 leads Qwen2.5 72B Instruct on the Artificial Analysis Intelligence Index, 26.1 to 7.7, and Qwen2.5 72B Instruct costs 11.0x less per output token ($0.40 vs $4.40 per million). Updated October 21, 2026; 37 benchmarks have a result for both.

Replace GLM 5.1 Replace Qwen2.5 72B Instruct
GLM 5.1Z.ai
Coding
49.2
Overall
57.5
Output $/M
$4.40
37
Qwen2.5 72B InstructQwen
Coding
No data
Overall
26.5
Output $/M
$0.40
0
vs boards won of 37 shared

GLM 5.1 or Qwen2.5 72B Instruct: which to pick

Only GLM 5.1 has a coding score, 49.2 on the Vibe Coding Index; Qwen2.5 72B Instruct has none, so the two cannot be ranked on coding.

Pick Qwen2.5 72B Instruct for output tokens 11.0x cheaper ($0.40 vs $4.40 per million).

Model AGLM 5.1Z.ai49.2/1003 independent benchmarks$0.60 per solved taskModel BQwen2.5 72B InstructQwenNot benchmarkedNo coding benchmarkNo cost per solved task
Shared results
37
GLM 5.1 only
52
Qwen2.5 72B Instruct only
16
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
GLM 5.1 49.2 from 3 benchmarks, Qwen2.5 72B Instruct not benchmarked.
Widest category
Agentic work, 63.2 points apart on average over 1 shared benchmark.
Widest single gap
OTIS Mock AIME: GLM 5.1 93.3% against Qwen2.5 72B Instruct 8.1%, 85.3 points apart on a 0 to 100 scale.
Output price
GLM 5.1 $4.40 against Qwen2.5 72B Instruct $0.40 per million output tokens, 11.0x apart.
Context window
GLM 5.1 takes 205K tokens, Qwen2.5 72B Instruct 33K.

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
CategoryGLM 5.1BarsQwen2.5 72B InstructTies
Coding40
Intelligence20
Agentic work10
Reasoning60
Math30
Knowledge40
General170
All shared boards370
CodingLike-for-likeSix of these boards feed the Vibe Coding Index40
Coding benchmarks for GLM 5.1 and Qwen2.5 72B Instruct
BenchmarkGLM 5.1Qwen2.5 72B InstructLeader
LMArena WebDev1508Not measuredOnly GLM 5.1 measured
AA Coding Index55.8Not measuredOnly GLM 5.1 measured
LiveCodeBench (Vals AI)81.4%Not measuredOnly GLM 5.1 measured
SWE-bench Verified (Vals AI)76.4%Not measuredOnly GLM 5.1 measured
Vibe Code Bench (Vals AI)31.5%Not measuredOnly GLM 5.1 measured
Code migration (Vals AI)25.8%Not measuredOnly GLM 5.1 measured
ProgramBench (Vals AI)0.0%Not measuredOnly GLM 5.1 measured
Terminal-Bench 2.1 (Vals AI)56.9%Not measuredOnly GLM 5.1 measured
LiveCodeBench (Artificial Analysis run)Not measured27.6%Only Qwen2.5 72B Instruct measured
SciCode44.8%Not measuredOnly GLM 5.1 measured
Terminal-Bench Hard (Artificial Analysis run)43.2%4.5%GLM 5.1
WeirdML57.1%16.0%GLM 5.1
ALE-Bench887.1Not measuredOnly GLM 5.1 measured
LMArena Text: coding14851292GLM 5.1
LMArena Text: software and IT14841292GLM 5.1
LMArena WebDev: React1505Not measuredOnly GLM 5.1 measured
LMArena WebDev: HTML1508Not measuredOnly GLM 5.1 measured
Neither model has a result on the other 22 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for GLM 5.1 and Qwen2.5 72B Instruct
BenchmarkGLM 5.1Qwen2.5 72B InstructLeader
AA Intelligence Index26.17.7GLM 5.1
Epoch Capabilities Index149.8129.0GLM 5.1
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
Agentic work benchmarks for GLM 5.1 and Qwen2.5 72B Instruct
BenchmarkGLM 5.1Qwen2.5 72B InstructLeader
AA Agentic Index23.9Not measuredOnly GLM 5.1 measured
Design Arena: agents building web apps1175Not measuredOnly GLM 5.1 measured
Design Arena: agents building full-stack apps1167Not measuredOnly GLM 5.1 measured
Design Arena: agents building mobile apps1162Not measuredOnly GLM 5.1 measured
Design Arena: agents building Android apps1180Not measuredOnly GLM 5.1 measured
Design Arena: agents building Godot games1053Not measuredOnly GLM 5.1 measured
Design Arena: agents building games1164Not measuredOnly GLM 5.1 measured
Design Arena: HTML slides1170Not measuredOnly GLM 5.1 measured
Design Arena: PowerPoint slides1188Not measuredOnly GLM 5.1 measured
Design Arena: spreadsheets1315Not measuredOnly GLM 5.1 measured
Finance Agent (Vals AI)44.8%Not measuredOnly GLM 5.1 measured
Legal Research Bench (Vals AI)27.9%Not measuredOnly GLM 5.1 measured
τ²-Bench Telecom97.7%34.5%GLM 5.1
APEX-Agents40.9%Not measuredOnly GLM 5.1 measured
METR time horizonsNot measured35.8%Only Qwen2.5 72B Instruct measured
BALROGNot measured16.2%Only Qwen2.5 72B Instruct measured
TheAgentCompanyNot measured5.7%Only Qwen2.5 72B Instruct measured
Vending-Bench 2$5,634Not measuredOnly GLM 5.1 measured
Neither model has a result on the other 23 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index60
Reasoning benchmarks for GLM 5.1 and Qwen2.5 72B Instruct
BenchmarkGLM 5.1Qwen2.5 72B InstructLeader
GPQA Diamond84.5%Not measuredOnly GLM 5.1 measured
GPQA Diamond (Artificial Analysis run)86.8%49.1%GLM 5.1
Humanity's Last Exam (Artificial Analysis run)30.1%3.6%GLM 5.1
GPQA Diamond (Epoch AI run)89.9%49.1%GLM 5.1
SimpleBench55.1%Not measuredOnly GLM 5.1 measured
Chess puzzles20.0%Not measuredOnly GLM 5.1 measured
BIG-Bench HardNot measured79.8%Only Qwen2.5 72B Instruct measured
LMArena Text: hard prompts14731271GLM 5.1
LMArena Text: expert14781245GLM 5.1
LMArena Text: hard prompts in English14821279GLM 5.1
Neither model has a result on the other 12 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index30
Math benchmarks for GLM 5.1 and Qwen2.5 72B Instruct
BenchmarkGLM 5.1Qwen2.5 72B InstructLeader
AIME91.9%Not measuredOnly GLM 5.1 measured
AA Math IndexNot measured14.0Only Qwen2.5 72B Instruct measured
AIME 2025 (Artificial Analysis run)Not measured14.0%Only Qwen2.5 72B Instruct measured
OTIS Mock AIME93.3%8.1%GLM 5.1
FrontierMath, tiers 1 to 336.8%Not measuredOnly GLM 5.1 measured
ProofBench22.2%Not measuredOnly GLM 5.1 measured
MATH level 5Not measured63.2%Only Qwen2.5 72B Instruct measured
LMArena Text: math14731283GLM 5.1
LMArena Text: mathematical work14701290GLM 5.1
Neither model has a result on the other 4 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index40
Knowledge benchmarks for GLM 5.1 and Qwen2.5 72B Instruct
BenchmarkGLM 5.1Qwen2.5 72B InstructLeader
MMLU Pro86.9%Not measuredOnly GLM 5.1 measured
LegalBench (Vals AI)84.4%Not measuredOnly GLM 5.1 measured
CaseLaw (Vals AI)51.5%Not measuredOnly GLM 5.1 measured
CorpFin (Vals AI)64.5%Not measuredOnly GLM 5.1 measured
TaxEval (Vals AI)71.2%Not measuredOnly GLM 5.1 measured
MedCode (Vals AI)41.6%Not measuredOnly GLM 5.1 measured
MMLU Pro (Artificial Analysis run)Not measured72.0%Only Qwen2.5 72B Instruct measured
SimpleQA Verified34.0%Not measuredOnly GLM 5.1 measured
MMLUNot measured85.3%Only Qwen2.5 72B Instruct measured
AI2 Reasoning ChallengeNot measured94.5%Only Qwen2.5 72B Instruct measured
LMArena Text: science14781275GLM 5.1
LMArena Text: medicine and healthcare14741258GLM 5.1
LMArena Text: legal and government14651288GLM 5.1
LMArena Text: business and finance14521253GLM 5.1
Neither model has a result on the other 4 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
Web design benchmarks for GLM 5.1 and Qwen2.5 72B Instruct
BenchmarkGLM 5.1Qwen2.5 72B InstructLeader
Design Arena: websites1290Not measuredOnly GLM 5.1 measured
Design Arena: UI components1292Not measuredOnly GLM 5.1 measured
Design Arena: game dev1265Not measuredOnly GLM 5.1 measured
Design Arena: data viz1366Not measuredOnly GLM 5.1 measured
Design Arena: 3D1336Not measuredOnly GLM 5.1 measured
Design Arena: all code categories1268Not measuredOnly GLM 5.1 measured
Design Arena: SVG1218Not measuredOnly GLM 5.1 measured
Design Arena: ASCII art1142Not measuredOnly GLM 5.1 measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index170
General benchmarks for GLM 5.1 and Qwen2.5 72B Instruct
BenchmarkGLM 5.1Qwen2.5 72B InstructLeader
Public Benefits Bench (Vals AI)61.8%Not measuredOnly GLM 5.1 measured
MedScribe (Vals AI)72.3%Not measuredOnly GLM 5.1 measured
IFBench76.3%36.9%GLM 5.1
Long context reasoning (AA-LCR)73.7%Not measuredOnly GLM 5.1 measured
DTBenchNot measured62.9%Only Qwen2.5 72B Instruct measured
LMCANot measured13.4%Only Qwen2.5 72B Instruct measured
ExploitBench18.1%Not measuredOnly GLM 5.1 measured
HellaSwagNot measured84.8%Only Qwen2.5 72B Instruct measured
WinoGrandeNot measured82.3%Only Qwen2.5 72B Instruct measured
PIQANot measured82.6%Only Qwen2.5 72B Instruct measured
LMArena Text14631269GLM 5.1
LMArena Text: instruction following14531254GLM 5.1
LMArena Text: creative writing14531221GLM 5.1
LMArena Text: English14761283GLM 5.1
LMArena Text: non-English14481252GLM 5.1
LMArena Text: Chinese15151273GLM 5.1
LMArena Text: German14671234GLM 5.1
LMArena Text: Japanese14351180GLM 5.1
LMArena Text: Korean14241188GLM 5.1
LMArena Text: French14701280GLM 5.1
LMArena Text: Spanish14681255GLM 5.1
LMArena Text: Russian14541264GLM 5.1
LMArena Text: Polish1449Not measuredOnly GLM 5.1 measured
LMArena Text: multi-turn14731272GLM 5.1
LMArena Text: longer queries14671281GLM 5.1
LMArena Text: writing and language14541247GLM 5.1
LMArena Text: entertainment and media14421211GLM 5.1
Neither model has a result on the other 19 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for GLM 5.1 and Qwen2.5 72B Instruct
BenchmarkGLM 5.1Qwen2.5 72B InstructLeader
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 5.1$0.0036Fits in one request
Qwen2.5 72B Instruct$0.0006Fits in one request

Qwen2.5 72B Instruct costs less

Repository review

50K input + 3K output tokens

GLM 5.1$0.083Fits in one request
Qwen2.5 72B Instruct$0.019Exceeds the context window

Qwen2.5 72B Instruct costs less

Agent loop

200K input + 10K output tokens

GLM 5.1$0.324Exceeds the context window
Qwen2.5 72B Instruct$0.076Exceeds the context window

Qwen2.5 72B Instruct 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 5.1 and Qwen2.5 72B Instruct
SpecificationZ.ai GLM 5.1Qwen2.5 72B Instruct
MakerZ.aiQwen
WeightsOpen weightsOpen weights
ReleasedApril 7, 2026September 19, 2024
Context window205K33K
Input $/Mtok$1.40$0.36
Output $/Mtok$4.40$0.40
Coding index49.2Not benchmarked
Independent measurements behind it30
Cost per solved task$0.60Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is GLM 5.1 better than Qwen2.5 72B Instruct for coding?

Only GLM 5.1 has a coding score, 49.2 on the Vibe Coding Index; Qwen2.5 72B Instruct has none, so the two cannot be ranked on coding. Pick Qwen2.5 72B Instruct for output tokens 11.0x cheaper ($0.40 vs $4.40 per million).

Which is cheaper, GLM 5.1 or Qwen2.5 72B Instruct?

GLM 5.1 costs $1.40 per million input tokens and $4.40 per million output tokens; Qwen2.5 72B Instruct costs $0.36 and $0.40. Cost per solved task is known for GLM 5.1 only.

Which has the bigger context window, GLM 5.1 or Qwen2.5 72B Instruct?

GLM 5.1, with 205K tokens against 33K.

How many benchmarks is the GLM 5.1 vs Qwen2.5 72B Instruct comparison based on?

37 benchmarks have a result for both models as of October 21, 2026. GLM 5.1's coding index rests on 3 independent benchmarks and Qwen2.5 72B Instruct's on 0. Where only one model has a result, the page says so instead of naming a winner.

Compare next