Enter opens the highlighted result. Esc closes.

DeepSeek V4 Pro 0423 vs Llama 3.3 70B Instruct

Updated October 21, 2026. 36 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace DeepSeek V4 Pro 0423 Replace Llama 3.3 70B Instruct
DeepSeek V4 Pro 0423DeepSeek
Coding
38.7
Overall
30.4
Output $/M
$2.716
36
Llama 3.3 70B InstructMeta Llama
Coding
No data
Overall
7.7
Output $/M
$0.72
0
vs boards won of 36 shared

Decision reading

Only DeepSeek V4 Pro 0423 has a coding score, 38.7 on the Vibe Coding Index; Llama 3.3 70B Instruct has none, so the two cannot be ranked on coding.

Llama 3.3 70B Instruct has the edge where output tokens cost 3.8x less.

Model ADeepSeek V4 Pro 0423DeepSeek38.7/1005 independent benchmarks$0.57 per solved taskModel BLlama 3.3 70B InstructMeta LlamaNot benchmarkedNo coding benchmarkNo cost per solved task
Shared results
36
DeepSeek V4 Pro 0423 only
47
Llama 3.3 70B Instruct only
15
Like-for-like categories
6 of 9

Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.

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 V4 Pro 0423 38.7 from 5 benchmarks, Llama 3.3 70B Instruct not benchmarked.
Widest category
Math, 50.7 points apart on average over 3 shared benchmarks.
Widest single gap
OTIS Mock AIME: DeepSeek V4 Pro 0423 96.7% against Llama 3.3 70B Instruct 5.1%, 91.5 points apart on a 0 to 100 scale.
Output price
DeepSeek V4 Pro 0423 $2.716 against Llama 3.3 70B Instruct $0.72 per million output tokens, 3.8x apart.
Context window
DeepSeek V4 Pro 0423 takes 1.05M tokens, Llama 3.3 70B 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
CategoryDeepSeek V4 Pro 0423BarsLlama 3.3 70B InstructTies
Coding40
Intelligence20
Reasoning40
Math30
Knowledge40
General190
All shared boards360
CodingLike-for-likeSix of these boards feed the Vibe Coding Index40
Coding benchmarks for DeepSeek V4 Pro 0423 and Llama 3.3 70B Instruct
BenchmarkDeepSeek V4 Pro 0423Llama 3.3 70B InstructLeader
LMArena WebDev1447Not measuredOnly DeepSeek V4 Pro 0423 measured
AA Coding Index59.411.9DeepSeek V4 Pro 0423
LiveCodeBench (Vals AI)87.5%Not measuredOnly DeepSeek V4 Pro 0423 measured
SWE-bench Verified (Vals AI)77.4%Not measuredOnly DeepSeek V4 Pro 0423 measured
Terminal-Bench 4 (Vals AI)11.1%Not measuredOnly DeepSeek V4 Pro 0423 measured
Vibe Code Bench (Vals AI)49.9%Not measuredOnly DeepSeek V4 Pro 0423 measured
Code migration (Vals AI)26.2%Not measuredOnly DeepSeek V4 Pro 0423 measured
ProgramBench (Vals AI)0.0%Not measuredOnly DeepSeek V4 Pro 0423 measured
Terminal-Bench 2.1 (Vals AI)50.2%Not measuredOnly DeepSeek V4 Pro 0423 measured
SkillsBench (Vals AI)51.3%Not measuredOnly DeepSeek V4 Pro 0423 measured
LiveCodeBench (Artificial Analysis run)Not measured28.8%Only Llama 3.3 70B Instruct measured
Terminal-Bench Hard (Artificial Analysis run)Not measured3.0%Only Llama 3.3 70B Instruct measured
FrontierCode17.6%Not measuredOnly DeepSeek V4 Pro 0423 measured
WeirdML48.9%14.4%DeepSeek V4 Pro 0423
ALE-Bench1006.1Not measuredOnly DeepSeek V4 Pro 0423 measured
LMArena Text: coding14701268DeepSeek V4 Pro 0423
LMArena Text: software and IT14691273DeepSeek V4 Pro 0423
LMArena WebDev: React1441Not measuredOnly DeepSeek V4 Pro 0423 measured
LMArena WebDev: HTML1435Not measuredOnly DeepSeek V4 Pro 0423 measured
Neither model has a result on the other 19 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for DeepSeek V4 Pro 0423 and Llama 3.3 70B Instruct
BenchmarkDeepSeek V4 Pro 0423Llama 3.3 70B InstructLeader
AA Intelligence Index30.47.7DeepSeek V4 Pro 0423
Epoch Capabilities Index149.1127.3DeepSeek V4 Pro 0423
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index11 boards, none shared
Agentic work benchmarks for DeepSeek V4 Pro 0423 and Llama 3.3 70B Instruct
BenchmarkDeepSeek V4 Pro 0423Llama 3.3 70B InstructLeader
AA Agentic Index26.3Not measuredOnly DeepSeek V4 Pro 0423 measured
Design Arena: agents building web apps1000Not measuredOnly DeepSeek V4 Pro 0423 measured
Design Arena: agents building full-stack apps948Not measuredOnly DeepSeek V4 Pro 0423 measured
Design Arena: agents building Godot games1059Not measuredOnly DeepSeek V4 Pro 0423 measured
Finance Agent (Vals AI)44.1%Not measuredOnly DeepSeek V4 Pro 0423 measured
Tax Agent Bench (Vals AI)25.2%Not measuredOnly DeepSeek V4 Pro 0423 measured
Legal Research Bench (Vals AI)23.1%Not measuredOnly DeepSeek V4 Pro 0423 measured
Excel Modeling Benchmark (Vals AI)51.6%Not measuredOnly DeepSeek V4 Pro 0423 measured
τ²-Bench TelecomNot measured26.6%Only Llama 3.3 70B Instruct measured
BALROGNot measured23.0%Only Llama 3.3 70B Instruct measured
Vending-Bench 2$3,285Not measuredOnly DeepSeek V4 Pro 0423 measured
Neither model has a result on the other 28 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index40
Reasoning benchmarks for DeepSeek V4 Pro 0423 and Llama 3.3 70B Instruct
BenchmarkDeepSeek V4 Pro 0423Llama 3.3 70B InstructLeader
GPQA Diamond89.4%Not measuredOnly DeepSeek V4 Pro 0423 measured
GPQA Diamond (Artificial Analysis run)Not measured49.8%Only Llama 3.3 70B Instruct measured
Humanity's Last Exam (Artificial Analysis run)Not measured3.6%Only Llama 3.3 70B Instruct measured
GPQA Diamond (Epoch AI run)90.9%47.4%DeepSeek V4 Pro 0423
SimpleBenchNot measured19.9%Only Llama 3.3 70B Instruct measured
Chess puzzles19.0%Not measuredOnly DeepSeek V4 Pro 0423 measured
Mystery game puzzles17.0%Not measuredOnly DeepSeek V4 Pro 0423 measured
LMArena Text: hard prompts14611257DeepSeek V4 Pro 0423
LMArena Text: expert14631225DeepSeek V4 Pro 0423
LMArena Text: hard prompts in English14661277DeepSeek V4 Pro 0423
Neither model has a result on the other 11 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index30
Math benchmarks for DeepSeek V4 Pro 0423 and Llama 3.3 70B Instruct
BenchmarkDeepSeek V4 Pro 0423Llama 3.3 70B InstructLeader
AA Math IndexNot measured7.7Only Llama 3.3 70B Instruct measured
AIME 2025 (Artificial Analysis run)Not measured7.7%Only Llama 3.3 70B Instruct measured
OTIS Mock AIME96.7%5.1%DeepSeek V4 Pro 0423
FrontierMath, tiers 1 to 345.3%Not measuredOnly DeepSeek V4 Pro 0423 measured
FrontierMath, tier 42.4%Not measuredOnly DeepSeek V4 Pro 0423 measured
ProofBench16.0%Not measuredOnly DeepSeek V4 Pro 0423 measured
MATH level 5Not measured41.6%Only Llama 3.3 70B Instruct measured
LMArena Text: math14431267DeepSeek V4 Pro 0423
LMArena Text: mathematical work14551267DeepSeek V4 Pro 0423
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 DeepSeek V4 Pro 0423 and Llama 3.3 70B Instruct
BenchmarkDeepSeek V4 Pro 0423Llama 3.3 70B InstructLeader
MMLU Pro87.3%Not measuredOnly DeepSeek V4 Pro 0423 measured
LegalBench (Vals AI)80.3%Not measuredOnly DeepSeek V4 Pro 0423 measured
CaseLaw (Vals AI)59.4%Not measuredOnly DeepSeek V4 Pro 0423 measured
CorpFin (Vals AI)61.4%Not measuredOnly DeepSeek V4 Pro 0423 measured
TaxEval (Vals AI)72.1%Not measuredOnly DeepSeek V4 Pro 0423 measured
MedCode (Vals AI)40.5%Not measuredOnly DeepSeek V4 Pro 0423 measured
MMLU Pro (Artificial Analysis run)Not measured71.3%Only Llama 3.3 70B Instruct measured
SimpleQA Verified47.0%Not measuredOnly DeepSeek V4 Pro 0423 measured
MMLUNot measured86.3%Only Llama 3.3 70B Instruct measured
LMArena Text: science14701284DeepSeek V4 Pro 0423
LMArena Text: medicine and healthcare14661275DeepSeek V4 Pro 0423
LMArena Text: legal and government14641284DeepSeek V4 Pro 0423
LMArena Text: business and finance14431243DeepSeek V4 Pro 0423
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 DeepSeek V4 Pro 0423 and Llama 3.3 70B Instruct
BenchmarkDeepSeek V4 Pro 0423Llama 3.3 70B InstructLeader
Design Arena: websites1235Not measuredOnly DeepSeek V4 Pro 0423 measured
Design Arena: UI components1222Not measuredOnly DeepSeek V4 Pro 0423 measured
Design Arena: game dev1229Not measuredOnly DeepSeek V4 Pro 0423 measured
Design Arena: data viz1202Not measuredOnly DeepSeek V4 Pro 0423 measured
Design Arena: 3D1254Not measuredOnly DeepSeek V4 Pro 0423 measured
Design Arena: all code categories1240Not measuredOnly DeepSeek V4 Pro 0423 measured
Design Arena: SVG1140Not measuredOnly DeepSeek V4 Pro 0423 measured
Design Arena: ASCII art1148Not measuredOnly DeepSeek V4 Pro 0423 measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index190
General benchmarks for DeepSeek V4 Pro 0423 and Llama 3.3 70B Instruct
BenchmarkDeepSeek V4 Pro 0423Llama 3.3 70B InstructLeader
Public Benefits Bench (Vals AI)62.9%Not measuredOnly DeepSeek V4 Pro 0423 measured
MedScribe (Vals AI)75.1%Not measuredOnly DeepSeek V4 Pro 0423 measured
IFBenchNot measured47.1%Only Llama 3.3 70B Instruct measured
Long context reasoning (AA-LCR)Not measured15.7%Only Llama 3.3 70B Instruct measured
DTBench90.7%59.5%DeepSeek V4 Pro 0423
LMCA41.2%17.5%DeepSeek V4 Pro 0423
Fiction.LiveBenchNot measured33.3%Only Llama 3.3 70B Instruct measured
Surface Evolver Bench40.0%Not measuredOnly DeepSeek V4 Pro 0423 measured
CL-bench Life13.5%Not measuredOnly DeepSeek V4 Pro 0423 measured
LMArena Text14511274DeepSeek V4 Pro 0423
LMArena Text: instruction following14451242DeepSeek V4 Pro 0423
LMArena Text: creative writing14461250DeepSeek V4 Pro 0423
LMArena Text: English14591303DeepSeek V4 Pro 0423
LMArena Text: non-English14391236DeepSeek V4 Pro 0423
LMArena Text: Chinese14811217DeepSeek V4 Pro 0423
LMArena Text: German14581251DeepSeek V4 Pro 0423
LMArena Text: Japanese14461150DeepSeek V4 Pro 0423
LMArena Text: Korean14091142DeepSeek V4 Pro 0423
LMArena Text: French14731282DeepSeek V4 Pro 0423
LMArena Text: Spanish14571269DeepSeek V4 Pro 0423
LMArena Text: Russian14531252DeepSeek V4 Pro 0423
LMArena Text: Polish14611259DeepSeek V4 Pro 0423
LMArena Text: multi-turn14681280DeepSeek V4 Pro 0423
LMArena Text: longer queries14581255DeepSeek V4 Pro 0423
LMArena Text: writing and language14471255DeepSeek V4 Pro 0423
LMArena Text: entertainment and media14311241DeepSeek V4 Pro 0423
Neither model has a result on the other 20 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for DeepSeek V4 Pro 0423 and Llama 3.3 70B Instruct
BenchmarkDeepSeek V4 Pro 0423Llama 3.3 70B 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

DeepSeek V4 Pro 0423$0.0026Fits in one request
Llama 3.3 70B Instruct$0.0008Fits in one request

Llama 3.3 70B Instruct costs less

Repository review

50K input + 3K output tokens

DeepSeek V4 Pro 0423$0.069Fits in one request
Llama 3.3 70B Instruct$0.025Fits in one request

Llama 3.3 70B Instruct costs less

Agent loop

200K input + 10K output tokens

DeepSeek V4 Pro 0423$0.272Fits in one request
Llama 3.3 70B Instruct$0.097Exceeds the context window

Llama 3.3 70B 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 DeepSeek V4 Pro 0423 and Llama 3.3 70B Instruct
SpecificationDeepSeek V4 Pro 0423Meta Llama 3.3 70B Instruct
MakerDeepSeekMeta Llama
WeightsOpen weightsOpen weights
ReleasedApril 24, 2026December 6, 2024
Context window1.05M131K
Input $/Mtok$1.225$0.45
Output $/Mtok$2.716$0.72
Coding index38.7Not benchmarked
Independent measurements behind it50
Cost per solved task$0.57Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is DeepSeek V4 Pro 0423 better than Llama 3.3 70B Instruct for coding?

Only DeepSeek V4 Pro 0423 has a coding score, 38.7 on the Vibe Coding Index; Llama 3.3 70B Instruct has none, so the two cannot be ranked on coding. Llama 3.3 70B Instruct has the edge where output tokens cost 3.8x less.

Which is cheaper, DeepSeek V4 Pro 0423 or Llama 3.3 70B Instruct?

DeepSeek V4 Pro 0423 costs $1.225 per million input tokens and $2.716 per million output tokens; Llama 3.3 70B Instruct costs $0.45 and $0.72. Cost per solved task is known for DeepSeek V4 Pro 0423 only.

Which has the bigger context window, DeepSeek V4 Pro 0423 or Llama 3.3 70B Instruct?

DeepSeek V4 Pro 0423, with 1.05M tokens against 131K.

How many benchmarks is the DeepSeek V4 Pro 0423 vs Llama 3.3 70B Instruct comparison based on?

36 benchmarks have a result for both models as of October 21, 2026. DeepSeek V4 Pro 0423's coding index rests on 5 independent benchmarks and Llama 3.3 70B Instruct's on 0. Where only one model has a result, the page says so instead of naming a winner.

Compare next