Enter opens the highlighted result. Esc closes.

GPT-4.1 vs Llama 3.2 3B Instruct

GPT-4.1 leads Llama 3.2 3B Instruct on the Artificial Analysis Intelligence Index, 12.7 to 5.7, and Llama 3.2 3B Instruct costs 24.1x less per output token ($0.3325 vs $8.00 per million). Updated October 21, 2026; 34 benchmarks have a result for both.

Replace GPT-4.1 Replace Llama 3.2 3B Instruct
GPT-4.1OpenAI
Coding
No data
Overall
36.5
Output $/M
$8.00
33
Llama 3.2 3B InstructMeta Llama
Coding
No data
Overall
No data
Output $/M
$0.3325
1
vs boards won of 34 shared

GPT-4.1 or Llama 3.2 3B Instruct: which to pick

Neither model has a coding score yet, so the two cannot be ranked on coding.

Pick GPT-4.1 for a larger context window (1.05M against 131K tokens) and work measured by Humanity's Last Exam (Artificial Analysis run), where it leads.

Model AGPT-4.1OpenAINot benchmarkedNo coding benchmark$0.37 per solved taskModel BLlama 3.2 3B InstructMeta LlamaNot benchmarkedNo coding benchmarkNo cost per solved task
Shared results
34
GPT-4.1 only
58
Llama 3.2 3B Instruct only
1
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
GPT-4.1 not benchmarked, Llama 3.2 3B Instruct not benchmarked.
Widest category
General, 48.4 points apart on average over 14 shared benchmarks.
Widest single gap
Long context reasoning (AA-LCR): GPT-4.1 68.3% against Llama 3.2 3B Instruct 4.3%, 64.0 points apart on a 0 to 100 scale.
Output price
GPT-4.1 $8.00 against Llama 3.2 3B Instruct $0.3325 per million output tokens, 24.1x apart.
Context window
GPT-4.1 takes 1.05M tokens, Llama 3.2 3B Instruct 131K.

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
CategoryGPT-4.1BarsLlama 3.2 3B InstructTies
Coding30
Intelligence10
Agentic work20
Reasoning41
Math40
Knowledge50
General140
All shared boards331
CodingLike-for-likeSix of these boards feed the Vibe Coding Index30
Coding benchmarks for GPT-4.1 and Llama 3.2 3B Instruct
BenchmarkGPT-4.1Llama 3.2 3B InstructLeader
Aider polyglot52.4%Not measuredOnly GPT-4.1 measured
SWE-bench Verified39.6%Not measuredOnly GPT-4.1 measured
SWE-bench Multimodal31.1%Not measuredOnly GPT-4.1 measured
LiveCodeBench (Vals AI)54.7%Not measuredOnly GPT-4.1 measured
LiveCodeBench (Artificial Analysis run)45.7%8.3%GPT-4.1
Terminal-Bench Hard (Artificial Analysis run)13.6%Not measuredOnly GPT-4.1 measured
WeirdML39.0%Not measuredOnly GPT-4.1 measured
ALE-Bench558.1Not measuredOnly GPT-4.1 measured
LMArena Text: coding13901098GPT-4.1
LMArena Text: software and IT13911103GPT-4.1
SciCode4.6%Not measuredOnly GPT-4.1 measured
Neither model has a result on the other 28 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
Intelligence benchmarks for GPT-4.1 and Llama 3.2 3B Instruct
BenchmarkGPT-4.1Llama 3.2 3B InstructLeader
AA Intelligence Index12.75.7GPT-4.1
Epoch Capabilities Index136.7Not measuredOnly GPT-4.1 measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index20
Agentic work benchmarks for GPT-4.1 and Llama 3.2 3B Instruct
BenchmarkGPT-4.1Llama 3.2 3B InstructLeader
τ²-Bench Telecom47.1%21.1%GPT-4.1
BALROGNot measured10.1%Only Llama 3.2 3B Instruct measured
Berkeley Function Calling Leaderboard (BFCL)54.0%21.9%GPT-4.1
BrowseComp-Plus36.4%Not measuredOnly GPT-4.1 measured
Neither model has a result on the other 37 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index41
Reasoning benchmarks for GPT-4.1 and Llama 3.2 3B Instruct
BenchmarkGPT-4.1Llama 3.2 3B InstructLeader
GPQA Diamond65.4%Not measuredOnly GPT-4.1 measured
MMMU Pro (Vals AI)72.4%Not measuredOnly GPT-4.1 measured
GPQA Diamond (Artificial Analysis run)66.6%25.5%GPT-4.1
Humanity's Last Exam (Artificial Analysis run)4.2%5.3%Llama 3.2 3B Instruct
GPQA Diamond (Epoch AI run)66.9%Not measuredOnly GPT-4.1 measured
ARC-AGI5.5%Not measuredOnly GPT-4.1 measured
ARC-AGI-20.4%Not measuredOnly GPT-4.1 measured
SimpleBench27.0%Not measuredOnly GPT-4.1 measured
Chess puzzles3.0%Not measuredOnly GPT-4.1 measured
LMArena Text: hard prompts13841095GPT-4.1
LMArena Text: expert13641090GPT-4.1
LMArena Text: hard prompts in English13931141GPT-4.1
LMArena Vision: diagrams1215Not measuredOnly GPT-4.1 measured
LMArena Vision: homework1248Not measuredOnly GPT-4.1 measured
Kagi LLM Benchmark52.3%Not measuredOnly GPT-4.1 measured
Neither model has a result on the other 7 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index40
Math benchmarks for GPT-4.1 and Llama 3.2 3B Instruct
BenchmarkGPT-4.1Llama 3.2 3B InstructLeader
AIME39.6%Not measuredOnly GPT-4.1 measured
MATH 500 (Vals AI)87.2%Not measuredOnly GPT-4.1 measured
MGSM (Vals AI)87.7%Not measuredOnly GPT-4.1 measured
AA Math Index34.73.3GPT-4.1
AIME 2025 (Artificial Analysis run)34.7%3.3%GPT-4.1
OTIS Mock AIME38.3%Not measuredOnly GPT-4.1 measured
FrontierMath, tiers 1 to 36.0%Not measuredOnly GPT-4.1 measured
MATH level 583.0%Not measuredOnly GPT-4.1 measured
LMArena Text: math13691126GPT-4.1
LMArena Text: mathematical work13601128GPT-4.1
Neither model has a result on the other 3 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index50
Knowledge benchmarks for GPT-4.1 and Llama 3.2 3B Instruct
BenchmarkGPT-4.1Llama 3.2 3B InstructLeader
MMLU Pro80.5%Not measuredOnly GPT-4.1 measured
LegalBench (Vals AI)83.1%Not measuredOnly GPT-4.1 measured
CaseLaw (Vals AI)69.9%Not measuredOnly GPT-4.1 measured
CorpFin (Vals AI)63.0%Not measuredOnly GPT-4.1 measured
TaxEval (Vals AI)75.1%Not measuredOnly GPT-4.1 measured
MortgageTax (Vals AI)65.9%Not measuredOnly GPT-4.1 measured
MedQA (Vals AI)91.2%Not measuredOnly GPT-4.1 measured
MMLU Pro (Artificial Analysis run)80.6%34.7%GPT-4.1
SimpleQA Verified31.1%Not measuredOnly GPT-4.1 measured
GeoBench72.0%Not measuredOnly GPT-4.1 measured
LMArena Text: science13821106GPT-4.1
LMArena Text: medicine and healthcare13791104GPT-4.1
LMArena Text: legal and government13991115GPT-4.1
LMArena Text: business and finance13691074GPT-4.1
FACTS Benchmark Suite45.8%Not measuredOnly GPT-4.1 measured
Neither model has a result on the other 3 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index6 boards, none shared
Web design benchmarks for GPT-4.1 and Llama 3.2 3B Instruct
BenchmarkGPT-4.1Llama 3.2 3B InstructLeader
Design Arena: websites1039Not measuredOnly GPT-4.1 measured
Design Arena: UI components1000Not measuredOnly GPT-4.1 measured
Design Arena: game dev1080Not measuredOnly GPT-4.1 measured
Design Arena: data viz1104Not measuredOnly GPT-4.1 measured
Design Arena: 3D850Not measuredOnly GPT-4.1 measured
Design Arena: all code categories1026Not measuredOnly GPT-4.1 measured
Neither model has a result on the other 3 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index140
General benchmarks for GPT-4.1 and Llama 3.2 3B Instruct
BenchmarkGPT-4.1Llama 3.2 3B InstructLeader
IFBench43.0%26.2%GPT-4.1
Long context reasoning (AA-LCR)68.3%4.3%GPT-4.1
DTBench68.3%Not measuredOnly GPT-4.1 measured
LMCA25.6%Not measuredOnly GPT-4.1 measured
Fiction.LiveBench63.9%Not measuredOnly GPT-4.1 measured
LMArena Text13831110GPT-4.1
LMArena Text: instruction following13671089GPT-4.1
LMArena Text: creative writing13631094GPT-4.1
LMArena Vision1210Not measuredOnly GPT-4.1 measured
LMArena Text: English13921180GPT-4.1
LMArena Text: non-English13701019GPT-4.1
LMArena Text: Chinese13821017GPT-4.1
LMArena Text: German13811056GPT-4.1
LMArena Text: Japanese1319Not measuredOnly GPT-4.1 measured
LMArena Text: Korean1338Not measuredOnly GPT-4.1 measured
LMArena Text: French1382Not measuredOnly GPT-4.1 measured
LMArena Text: Spanish1376Not measuredOnly GPT-4.1 measured
LMArena Text: Russian1377949GPT-4.1
LMArena Text: Polish1389Not measuredOnly GPT-4.1 measured
LMArena Text: multi-turn13971105GPT-4.1
LMArena Text: longer queries13851100GPT-4.1
LMArena Text: writing and language13711096GPT-4.1
LMArena Text: entertainment and media13651083GPT-4.1
LMArena Vision: English1209Not measuredOnly GPT-4.1 measured
LMArena Vision: Chinese1226Not measuredOnly GPT-4.1 measured
LMArena Vision: reading text1218Not measuredOnly GPT-4.1 measured
LMArena Vision: humor1196Not measuredOnly GPT-4.1 measured
LMArena Vision: captioning1207Not measuredOnly GPT-4.1 measured
LMArena Vision: recognizing things1208Not measuredOnly GPT-4.1 measured
LMArena Vision: creative writing1210Not measuredOnly GPT-4.1 measured
Neither model has a result on the other 16 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for GPT-4.1 and Llama 3.2 3B Instruct
BenchmarkGPT-4.1Llama 3.2 3B 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

GPT-4.1$0.0060Fits in one request
Llama 3.2 3B Instruct$0.0002Fits in one request

Llama 3.2 3B Instruct costs less

Repository review

50K input + 3K output tokens

GPT-4.1$0.124Fits in one request
Llama 3.2 3B Instruct$0.0035Fits in one request

Llama 3.2 3B Instruct costs less

Agent loop

200K input + 10K output tokens

GPT-4.1$0.480Fits in one request
Llama 3.2 3B Instruct$0.013Exceeds the context window

Llama 3.2 3B 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 GPT-4.1 and Llama 3.2 3B Instruct
SpecificationOpenAI GPT-4.1Meta Llama 3.2 3B Instruct
MakerOpenAIMeta Llama
WeightsNo public weights listedOpen weights
ReleasedApril 14, 2025September 25, 2024
Context window1.05M131K
Input $/Mtok$2.00$0.0505
Output $/Mtok$8.00$0.3325
Coding indexNot benchmarkedNot benchmarked
Independent measurements behind it00
Cost per solved task$0.37Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is GPT-4.1 better than Llama 3.2 3B Instruct for coding?

Neither model has a coding score yet, so the two cannot be ranked on coding. Pick GPT-4.1 for a larger context window (1.05M against 131K tokens) and work measured by Humanity's Last Exam (Artificial Analysis run), where it leads.

Which is cheaper, GPT-4.1 or Llama 3.2 3B Instruct?

GPT-4.1 costs $2.00 per million input tokens and $8.00 per million output tokens; Llama 3.2 3B Instruct costs $0.0505 and $0.3325. Cost per solved task is known for GPT-4.1 only.

Which has the bigger context window, GPT-4.1 or Llama 3.2 3B Instruct?

GPT-4.1, with 1.05M tokens against 131K.

How many benchmarks is the GPT-4.1 vs Llama 3.2 3B Instruct comparison based on?

34 benchmarks have a result for both models as of October 21, 2026. GPT-4.1's coding index rests on 0 independent benchmarks and Llama 3.2 3B Instruct's on 0. Where only one model has a result, the page says so instead of naming a winner.

Compare next