Enter opens the highlighted result. Esc closes.

DeepSeek V3.2 vs Llama 4 Maverick

DeepSeek V3.2 leads Llama 4 Maverick on the Artificial Analysis Intelligence Index, 21.5 to 9.3, at about the same output price ($0.88 vs $1.00 per million). Updated October 21, 2026; 61 benchmarks have a result for both.

Replace DeepSeek V3.2 Replace Llama 4 Maverick
DeepSeek V3.2DeepSeek
Coding
30.0
Overall
37.9
Output $/M
$0.88
59
Llama 4 MaverickMeta Llama
Coding
No data
Overall
21.8
Output $/M
$1.00
2
vs boards won of 61 shared

DeepSeek V3.2 or Llama 4 Maverick: which to pick

Only DeepSeek V3.2 has a coding score, 30.0 on the Vibe Coding Index; Llama 4 Maverick has none, so the two cannot be ranked on coding.

Pick Llama 4 Maverick for a larger context window (524K against 164K tokens) and work measured by MGSM (Vals AI) or LegalBench (Vals AI), where it leads.

Model ADeepSeek V3.2DeepSeek30.0/1003 independent benchmarks$0.99 per solved taskModel BLlama 4 MaverickMeta LlamaNot benchmarkedNo coding benchmark$1.491 per solved task
Shared results
61
DeepSeek V3.2 only
14
Llama 4 Maverick only
24
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.

At a glance

Coding index
DeepSeek V3.2 30.0 from 3 benchmarks, Llama 4 Maverick not benchmarked.
Widest category
Agentic work, 72.8 points apart on average over 1 shared benchmark.
Widest single gap
τ²-Bench Telecom: DeepSeek V3.2 90.6% against Llama 4 Maverick 17.8%, 72.8 points apart on a 0 to 100 scale.
Output price
DeepSeek V3.2 $0.88 against Llama 4 Maverick $1.00 per million output tokens, 1.1x apart.
Context window
DeepSeek V3.2 takes 164K tokens, Llama 4 Maverick 524K.

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
CategoryDeepSeek V3.2BarsLlama 4 MaverickTies
Coding70
Intelligence20
Agentic work10
Reasoning80
Math51
Knowledge91
Web design60
General210
All shared boards592
CodingLike-for-likeSix of these boards feed the Vibe Coding Index70
Coding benchmarks for DeepSeek V3.2 and Llama 4 Maverick
BenchmarkDeepSeek V3.2Llama 4 MaverickLeader
Aider polyglotNot measured15.6%Only Llama 4 Maverick measured
SWE-bench Verified70.0%21.0%DeepSeek V3.2
SWE-bench Multilingual59.0%Not measuredOnly DeepSeek V3.2 measured
LMArena WebDev1362Not measuredOnly DeepSeek V3.2 measured
AA Coding Index44.216.3DeepSeek V3.2
LiveCodeBench (Vals AI)80.7%47.3%DeepSeek V3.2
SWE-bench Verified (Vals AI)67.6%Not measuredOnly DeepSeek V3.2 measured
Vibe Code Bench (Vals AI)5.1%Not measuredOnly DeepSeek V3.2 measured
LiveCodeBench (Artificial Analysis run)86.2%39.7%DeepSeek V3.2
SciCodeNot measured31.7%Only Llama 4 Maverick measured
Terminal-Bench Hard (Artificial Analysis run)35.6%6.8%DeepSeek V3.2
Terminal-Bench39.6%Not measuredOnly DeepSeek V3.2 measured
WeirdMLNot measured24.5%Only Llama 4 Maverick measured
ALE-BenchNot measured173.0Only Llama 4 Maverick measured
LMArena Text: coding14531301DeepSeek V3.2
LMArena Text: software and IT14491299DeepSeek V3.2
LMArena WebDev: React1350Not measuredOnly DeepSeek V3.2 measured
LMArena WebDev: HTML1371Not measuredOnly DeepSeek V3.2 measured
Neither model has a result on the other 21 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for DeepSeek V3.2 and Llama 4 Maverick
BenchmarkDeepSeek V3.2Llama 4 MaverickLeader
AA Intelligence Index21.59.3DeepSeek V3.2
Epoch Capabilities Index146.1132.2DeepSeek V3.2
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
Agentic work benchmarks for DeepSeek V3.2 and Llama 4 Maverick
BenchmarkDeepSeek V3.2Llama 4 MaverickLeader
AA Agentic IndexNot measured0.6Only Llama 4 Maverick measured
τ²-Bench Telecom90.6%17.8%DeepSeek V3.2
APEX-Agents21.3%Not measuredOnly DeepSeek V3.2 measured
Vending-Bench 2$1,034Not measuredOnly DeepSeek V3.2 measured
Neither model has a result on the other 37 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index80
Reasoning benchmarks for DeepSeek V3.2 and Llama 4 Maverick
BenchmarkDeepSeek V3.2Llama 4 MaverickLeader
GPQA Diamond80.3%69.4%DeepSeek V3.2
MMMU Pro (Vals AI)Not measured71.7%Only Llama 4 Maverick measured
GPQA Diamond (Artificial Analysis run)84.0%67.1%DeepSeek V3.2
Humanity's Last Exam (Artificial Analysis run)24.6%4.9%DeepSeek V3.2
ARC-AGI57.0%4.4%DeepSeek V3.2
ARC-AGI-24.0%0.0%DeepSeek V3.2
SimpleBenchNot measured27.7%Only Llama 4 Maverick measured
LMArena Text: hard prompts14341281DeepSeek V3.2
LMArena Text: expert14361259DeepSeek V3.2
LMArena Text: hard prompts in English14461295DeepSeek V3.2
LMArena Vision: diagramsNot measured1134Only Llama 4 Maverick measured
LMArena Vision: homeworkNot measured1164Only Llama 4 Maverick measured
Kagi LLM BenchmarkNot measured55.9%Only Llama 4 Maverick measured
Neither model has a result on the other 9 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index51
Math benchmarks for DeepSeek V3.2 and Llama 4 Maverick
BenchmarkDeepSeek V3.2Llama 4 MaverickLeader
AIME84.6%25.2%DeepSeek V3.2
MATH 500 (Vals AI)Not measured85.2%Only Llama 4 Maverick measured
MGSM (Vals AI)86.0%92.4%Llama 4 Maverick
AA Math Index92.019.3DeepSeek V3.2
AIME 2025 (Artificial Analysis run)92.0%19.3%DeepSeek V3.2
LMArena Text: math14351299DeepSeek V3.2
LMArena Text: mathematical work14401295DeepSeek V3.2
Neither model has a result on the other 6 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index91
Knowledge benchmarks for DeepSeek V3.2 and Llama 4 Maverick
BenchmarkDeepSeek V3.2Llama 4 MaverickLeader
MMLU Pro84.9%79.4%DeepSeek V3.2
LegalBench (Vals AI)76.1%77.8%Llama 4 Maverick
CaseLaw (Vals AI)55.4%Not measuredOnly DeepSeek V3.2 measured
CorpFin (Vals AI)51.0%49.7%DeepSeek V3.2
TaxEval (Vals AI)68.2%66.6%DeepSeek V3.2
MortgageTax (Vals AI)Not measured58.5%Only Llama 4 Maverick measured
MedQA (Vals AI)93.9%43.3%DeepSeek V3.2
MedCode (Vals AI)Not measured36.5%Only Llama 4 Maverick measured
MMLU Pro (Artificial Analysis run)86.2%80.9%DeepSeek V3.2
GeoBenchNot measured52.0%Only Llama 4 Maverick measured
LMArena Text: science14461288DeepSeek V3.2
LMArena Text: medicine and healthcare14401275DeepSeek V3.2
LMArena Text: legal and government14251286DeepSeek V3.2
LMArena Text: business and finance14101260DeepSeek V3.2
Neither model has a result on the other 4 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index60
Web design benchmarks for DeepSeek V3.2 and Llama 4 Maverick
BenchmarkDeepSeek V3.2Llama 4 MaverickLeader
Design Arena: websites1175871DeepSeek V3.2
Design Arena: UI components1149899DeepSeek V3.2
Design Arena: game dev1132838DeepSeek V3.2
Design Arena: data viz1158882DeepSeek V3.2
Design Arena: 3D1133901DeepSeek V3.2
Design Arena: all code categories1164880DeepSeek V3.2
Design Arena: SVG1032Not measuredOnly DeepSeek V3.2 measured
Design Arena: ASCII art1083Not measuredOnly DeepSeek V3.2 measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index210
General benchmarks for DeepSeek V3.2 and Llama 4 Maverick
BenchmarkDeepSeek V3.2Llama 4 MaverickLeader
SAGE (Vals AI)Not measured35.7%Only Llama 4 Maverick measured
MedScribe (Vals AI)Not measured54.2%Only Llama 4 Maverick measured
IFBench60.7%43.0%DeepSeek V3.2
Long context reasoning (AA-LCR)73.3%50.0%DeepSeek V3.2
DTBench85.6%61.9%DeepSeek V3.2
LMCA28.8%15.9%DeepSeek V3.2
Fiction.LiveBenchNot measured46.2%Only Llama 4 Maverick measured
Creative writing (Lech Mazur)Not measured62.0%Only Llama 4 Maverick measured
CL-bench12.4%Not measuredOnly DeepSeek V3.2 measured
CL-bench Life7.4%Not measuredOnly DeepSeek V3.2 measured
LMArena Text14251287DeepSeek V3.2
LMArena Text: instruction following14131267DeepSeek V3.2
LMArena Text: creative writing14001266DeepSeek V3.2
LMArena VisionNot measured1142Only Llama 4 Maverick measured
LMArena Text: English14391303DeepSeek V3.2
LMArena Text: non-English14091269DeepSeek V3.2
LMArena Text: Chinese14551277DeepSeek V3.2
LMArena Text: German14131291DeepSeek V3.2
LMArena Text: Japanese13741206DeepSeek V3.2
LMArena Text: Korean13701203DeepSeek V3.2
LMArena Text: French14291259DeepSeek V3.2
LMArena Text: Spanish14241293DeepSeek V3.2
LMArena Text: Russian14241286DeepSeek V3.2
LMArena Text: Polish14191275DeepSeek V3.2
LMArena Text: multi-turn14271289DeepSeek V3.2
LMArena Text: longer queries14281280DeepSeek V3.2
LMArena Text: writing and language14071277DeepSeek V3.2
LMArena Text: entertainment and media13951250DeepSeek V3.2
LMArena Vision: EnglishNot measured1150Only Llama 4 Maverick measured
LMArena Vision: ChineseNot measured1141Only Llama 4 Maverick measured
LMArena Vision: reading textNot measured1143Only Llama 4 Maverick measured
LMArena Vision: humorNot measured1100Only Llama 4 Maverick measured
LMArena Vision: creative writingNot measured1099Only Llama 4 Maverick measured
Neither model has a result on the other 13 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for DeepSeek V3.2 and Llama 4 Maverick
BenchmarkDeepSeek V3.2Llama 4 MaverickLeader
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

DeepSeek V3.2$0.0007Fits in one request
Llama 4 Maverick$0.0008Fits in one request

DeepSeek V3.2 costs less

Repository review

50K input + 3K output tokens

DeepSeek V3.2$0.017Fits in one request
Llama 4 Maverick$0.021Fits in one request

DeepSeek V3.2 costs less

Agent loop

200K input + 10K output tokens

DeepSeek V3.2$0.067Exceeds the context window
Llama 4 Maverick$0.080Fits in one request

DeepSeek V3.2 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 DeepSeek V3.2 and Llama 4 Maverick
SpecificationDeepSeek V3.2Meta Llama 4 Maverick
MakerDeepSeekMeta Llama
WeightsOpen weightsOpen weights
ReleasedDecember 1, 2025April 5, 2025
Context window164K524K
Input $/Mtok$0.29$0.35
Output $/Mtok$0.88$1.00
Coding index30.0Not benchmarked
Independent measurements behind it30
Cost per solved task$0.99$1.491
SourceOpenRouter listingOpenRouter listing

Questions

Is DeepSeek V3.2 better than Llama 4 Maverick for coding?

Only DeepSeek V3.2 has a coding score, 30.0 on the Vibe Coding Index; Llama 4 Maverick has none, so the two cannot be ranked on coding. Pick Llama 4 Maverick for a larger context window (524K against 164K tokens) and work measured by MGSM (Vals AI) or LegalBench (Vals AI), where it leads.

Which is cheaper, DeepSeek V3.2 or Llama 4 Maverick?

DeepSeek V3.2 costs $0.29 per million input tokens and $0.88 per million output tokens; Llama 4 Maverick costs $0.35 and $1.00. Per solved SWE-bench Verified task, DeepSeek V3.2 costs $0.99 and Llama 4 Maverick $1.491.

Which has the bigger context window, DeepSeek V3.2 or Llama 4 Maverick?

Llama 4 Maverick, with 524K tokens against 164K.

How many benchmarks is the DeepSeek V3.2 vs Llama 4 Maverick comparison based on?

61 benchmarks have a result for both models as of October 21, 2026. DeepSeek V3.2's coding index rests on 3 independent benchmarks and Llama 4 Maverick's on 0. Where only one model has a result, the page says so instead of naming a winner.

Compare next