Enter opens the highlighted result. Esc closes.

GLM 5 vs Mistral Small 3

GLM 5 leads Mistral Small 3 on the Artificial Analysis Intelligence Index, 27.9 to 6.7, and Mistral Small 3 costs 40.0x less per output token ($0.08 vs $3.20 per million). Updated October 21, 2026; 32 benchmarks have a result for both.

Replace GLM 5 Replace Mistral Small 3
GLM 5Z.ai
Coding
41.7
Overall
27.9
Output $/M
$3.20
32
Mistral Small 3Mistral AI
Coding
No data
Overall
6.7
Output $/M
$0.08
0
vs boards won of 32 shared

GLM 5 or Mistral Small 3: which to pick

Only GLM 5 has a coding score, 41.7 on the Vibe Coding Index; Mistral Small 3 has none, so the two cannot be ranked on coding.

Pick Mistral Small 3 for output tokens 40.0x cheaper ($0.08 vs $3.20 per million).

Model AGLM 5Z.ai41.7/1003 independent benchmarks$0.63 per solved taskModel BMistral Small 3Mistral AINot benchmarkedNo coding benchmarkNo cost per solved task
Shared results
32
GLM 5 only
43
Mistral Small 3 only
4
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 41.7 from 3 benchmarks, Mistral Small 3 not benchmarked.
Widest category
Agentic work, 78.7 points apart on average over 1 shared benchmark.
Widest single gap
τ²-Bench Telecom: GLM 5 98.2% against Mistral Small 3 19.6%, 78.7 points apart on a 0 to 100 scale.
Output price
GLM 5 $3.20 against Mistral Small 3 $0.08 per million output tokens, 40.0x apart.
Context window
GLM 5 takes 205K tokens, Mistral Small 3 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 5BarsMistral Small 3Ties
Coding20
Intelligence10
Agentic work10
Reasoning50
Math20
Knowledge40
General170
All shared boards320
CodingLike-for-likeSix of these boards feed the Vibe Coding Index20
Coding benchmarks for GLM 5 and Mistral Small 3
BenchmarkGLM 5Mistral Small 3Leader
SWE-bench Verified72.8%Not measuredOnly GLM 5 measured
SWE-bench Multilingual69.7%Not measuredOnly GLM 5 measured
LMArena WebDev1435Not measuredOnly GLM 5 measured
LiveCodeBench (Vals AI)81.9%Not measuredOnly GLM 5 measured
SWE-bench Verified (Vals AI)71.4%Not measuredOnly GLM 5 measured
Vibe Code Bench (Vals AI)23.4%Not measuredOnly GLM 5 measured
LiveCodeBench (Artificial Analysis run)Not measured25.2%Only Mistral Small 3 measured
Terminal-Bench Hard (Artificial Analysis run)43.2%Not measuredOnly GLM 5 measured
Terminal-Bench52.4%Not measuredOnly GLM 5 measured
WeirdML48.2%Not measuredOnly GLM 5 measured
ALE-Bench765.6Not measuredOnly GLM 5 measured
LMArena Text: coding14611246GLM 5
LMArena Text: software and IT14601248GLM 5
LMArena WebDev: React1424Not measuredOnly GLM 5 measured
LMArena WebDev: HTML1450Not measuredOnly GLM 5 measured
Neither model has a result on the other 23 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
Intelligence benchmarks for GLM 5 and Mistral Small 3
BenchmarkGLM 5Mistral Small 3Leader
AA Intelligence Index27.96.7GLM 5
Epoch Capabilities Index145.8Not measuredOnly GLM 5 measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
Agentic work benchmarks for GLM 5 and Mistral Small 3
BenchmarkGLM 5Mistral Small 3Leader
Design Arena: agents building full-stack apps1123Not measuredOnly GLM 5 measured
Design Arena: agents building mobile apps1136Not measuredOnly GLM 5 measured
Design Arena: agents building Android apps1146Not measuredOnly GLM 5 measured
Design Arena: agents building Godot games1144Not measuredOnly GLM 5 measured
Design Arena: HTML slides1148Not measuredOnly GLM 5 measured
τ²-Bench Telecom98.2%19.6%GLM 5
Vending-Bench 2$4,432Not measuredOnly GLM 5 measured
Neither model has a result on the other 32 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index50
Reasoning benchmarks for GLM 5 and Mistral Small 3
BenchmarkGLM 5Mistral Small 3Leader
GPQA Diamond83.3%Not measuredOnly GLM 5 measured
GPQA Diamond (Artificial Analysis run)82.0%46.2%GLM 5
Humanity's Last Exam (Artificial Analysis run)29.3%3.8%GLM 5
GPQA Diamond (Epoch AI run)87.8%Not measuredOnly GLM 5 measured
ARC-AGI44.7%Not measuredOnly GLM 5 measured
ARC-AGI-24.9%Not measuredOnly GLM 5 measured
SimpleBench53.2%Not measuredOnly GLM 5 measured
LMArena Text: hard prompts14521233GLM 5
LMArena Text: expert14541202GLM 5
LMArena Text: hard prompts in English14601243GLM 5
Neither model has a result on the other 11 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index20
Math benchmarks for GLM 5 and Mistral Small 3
BenchmarkGLM 5Mistral Small 3Leader
AIME91.7%Not measuredOnly GLM 5 measured
AA Math IndexNot measured4.3Only Mistral Small 3 measured
AIME 2025 (Artificial Analysis run)Not measured4.3%Only Mistral Small 3 measured
OTIS Mock AIME80.0%Not measuredOnly GLM 5 measured
LMArena Text: math14401240GLM 5
LMArena Text: mathematical work14511243GLM 5
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index40
Knowledge benchmarks for GLM 5 and Mistral Small 3
BenchmarkGLM 5Mistral Small 3Leader
MMLU Pro86.0%Not measuredOnly GLM 5 measured
LegalBench (Vals AI)84.1%Not measuredOnly GLM 5 measured
CaseLaw (Vals AI)52.5%Not measuredOnly GLM 5 measured
CorpFin (Vals AI)62.9%Not measuredOnly GLM 5 measured
TaxEval (Vals AI)70.0%Not measuredOnly GLM 5 measured
MedQA (Vals AI)94.3%Not measuredOnly GLM 5 measured
MMLU Pro (Artificial Analysis run)Not measured65.2%Only Mistral Small 3 measured
LMArena Text: science14601235GLM 5
LMArena Text: medicine and healthcare14551215GLM 5
LMArena Text: legal and government14531219GLM 5
LMArena Text: business and finance14301205GLM 5
Neither model has a result on the other 6 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
Web design benchmarks for GLM 5 and Mistral Small 3
BenchmarkGLM 5Mistral Small 3Leader
Design Arena: websites1247Not measuredOnly GLM 5 measured
Design Arena: UI components1230Not measuredOnly GLM 5 measured
Design Arena: game dev1234Not measuredOnly GLM 5 measured
Design Arena: data viz1228Not measuredOnly GLM 5 measured
Design Arena: 3D1229Not measuredOnly GLM 5 measured
Design Arena: all code categories1246Not measuredOnly GLM 5 measured
Design Arena: SVG1168Not measuredOnly GLM 5 measured
Design Arena: ASCII art1147Not measuredOnly GLM 5 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 and Mistral Small 3
BenchmarkGLM 5Mistral Small 3Leader
IFBench72.3%26.4%GLM 5
Long context reasoning (AA-LCR)75.7%0.0%GLM 5
CL-bench18.7%Not measuredOnly GLM 5 measured
LMArena Text14461234GLM 5
LMArena Text: instruction following14281214GLM 5
LMArena Text: creative writing14381194GLM 5
LMArena Text: English14581256GLM 5
LMArena Text: non-English14301198GLM 5
LMArena Text: Chinese15111204GLM 5
LMArena Text: German14471211GLM 5
LMArena Text: Japanese14161111GLM 5
LMArena Text: Korean14231188GLM 5
LMArena Text: French14561204GLM 5
LMArena Text: Spanish1454Not measuredOnly GLM 5 measured
LMArena Text: Russian14361216GLM 5
LMArena Text: Polish1434Not measuredOnly GLM 5 measured
LMArena Text: multi-turn14561217GLM 5
LMArena Text: longer queries14461246GLM 5
LMArena Text: writing and language14371214GLM 5
LMArena Text: entertainment and media14261184GLM 5
Neither model has a result on the other 26 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for GLM 5 and Mistral Small 3
BenchmarkGLM 5Mistral Small 3Leader
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$0.0026Fits in one request
Mistral Small 3$0.0001Fits in one request

Mistral Small 3 costs less

Repository review

50K input + 3K output tokens

GLM 5$0.060Fits in one request
Mistral Small 3$0.0027Exceeds the context window

Mistral Small 3 costs less

Agent loop

200K input + 10K output tokens

GLM 5$0.232Exceeds the context window
Mistral Small 3$0.011Exceeds the context window

Mistral Small 3 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 and Mistral Small 3
SpecificationZ.ai GLM 5Mistral Small 3
MakerZ.aiMistral AI
WeightsOpen weightsOpen weights
ReleasedFebruary 11, 2026January 30, 2025
Context window205K33K
Input $/Mtok$1.00$0.05
Output $/Mtok$3.20$0.08
Coding index41.7Not benchmarked
Independent measurements behind it30
Cost per solved task$0.63Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is GLM 5 better than Mistral Small 3 for coding?

Only GLM 5 has a coding score, 41.7 on the Vibe Coding Index; Mistral Small 3 has none, so the two cannot be ranked on coding. Pick Mistral Small 3 for output tokens 40.0x cheaper ($0.08 vs $3.20 per million).

Which is cheaper, GLM 5 or Mistral Small 3?

GLM 5 costs $1.00 per million input tokens and $3.20 per million output tokens; Mistral Small 3 costs $0.05 and $0.08. Cost per solved task is known for GLM 5 only.

Which has the bigger context window, GLM 5 or Mistral Small 3?

GLM 5, with 205K tokens against 33K.

How many benchmarks is the GLM 5 vs Mistral Small 3 comparison based on?

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

Compare next