LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

DeepSeek V4 Pro 0813 vs MiniMax M2.7

Updated October 21, 2026. 56 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 0813 Replace MiniMax M2.7
DeepSeek V4 Pro 0813DeepSeek
Coding
54.6
Overall
36.0
Output $/M
$1.98
54
MiniMax M2.7MiniMax
Coding
28.4
Overall
22.8
Output $/M
$1.20
1
vs boards won of 56 shared, 1 tied

Decision reading

Pick DeepSeek V4 Pro 0813 for coding: it scores 54.6 on the Vibe Coding Index against 28.4 for MiniMax M2.7.

Pick MiniMax M2.7 when output tokens cost 1.7x less, or when the benchmark that matters is LegalBench (Vals AI), where it leads.

Model ADeepSeek V4 Pro 0813DeepSeek54.6/1001 independent benchmark$0.10 per solved taskModel BMiniMax M2.7MiniMax28.4/1001 independent benchmark$0.64 per solved task
Shared results
56
DeepSeek V4 Pro 0813 only
22
MiniMax M2.7 only
18
Like-for-like categories
7 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 0813 54.6 from 1 benchmark, MiniMax M2.7 28.4 from 1 benchmark.
Widest category
Agentic work, 24.6 points apart on average over 5 shared benchmarks.
Widest single gap
Vibe Code Bench (Vals AI): DeepSeek V4 Pro 0813 82.3% against MiniMax M2.7 11.9%, 70.4 points apart on a 0 to 100 scale.
Output price
DeepSeek V4 Pro 0813 $1.98 against MiniMax M2.7 $1.20 per million output tokens, 1.7x apart.
Context window
DeepSeek V4 Pro 0813 takes 1.05M tokens, MiniMax M2.7 205K.

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 0813BarsMiniMax M2.7Ties
Coding1401 tie
Intelligence20
Agentic work50
Reasoning60
Math30
Knowledge81
General160
All shared boards5411 tie
CodingLike-for-likeFour of these boards feed the Vibe Coding Index140
Coding benchmarks for DeepSeek V4 Pro 0813 and MiniMax M2.7
BenchmarkDeepSeek V4 Pro 0813MiniMax M2.7Leader
LMArena WebDev15821399DeepSeek V4 Pro 0813
AA Coding Index68.852.6DeepSeek V4 Pro 0813
LiveCodeBench (Vals AI)87.5%79.9%DeepSeek V4 Pro 0813
SWE-bench Verified (Vals AI)96.4%73.8%DeepSeek V4 Pro 0813
Terminal-Bench 4 (Vals AI)14.1%Not measuredOnly DeepSeek V4 Pro 0813 measured
Vibe Code Bench (Vals AI)82.3%11.9%DeepSeek V4 Pro 0813
Code migration (Vals AI)41.5%8.7%DeepSeek V4 Pro 0813
ProgramBench (Vals AI)0.0%0.0%Tie
IOI (Vals AI)51.6%Not measuredOnly DeepSeek V4 Pro 0813 measured
Terminal-Bench 2.1 (Vals AI)54.7%48.7%DeepSeek V4 Pro 0813
Vibe Code Bench 1-100 (Vals AI)17.5%Not measuredOnly DeepSeek V4 Pro 0813 measured
SkillsBench (Vals AI)53.8%Not measuredOnly DeepSeek V4 Pro 0813 measured
CyberBench (Vals AI)72.9%Not measuredOnly DeepSeek V4 Pro 0813 measured
SciCode51.0%50.1%DeepSeek V4 Pro 0813
Terminal-Bench Hard (Artificial Analysis run)Not measured39.4%Only MiniMax M2.7 measured
Terminal-BenchNot measured45.1%Only MiniMax M2.7 measured
FrontierCode28.5%Not measuredOnly DeepSeek V4 Pro 0813 measured
WeirdML66.2%37.0%DeepSeek V4 Pro 0813
ALE-Bench1403.2599.3DeepSeek V4 Pro 0813
LMArena Text: coding14691454DeepSeek V4 Pro 0813
LMArena Text: software and IT14621446DeepSeek V4 Pro 0813
LMArena WebDev: React15881388DeepSeek V4 Pro 0813
LMArena WebDev: HTML15661401DeepSeek V4 Pro 0813
Neither model has a result on the other 15 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for DeepSeek V4 Pro 0813 and MiniMax M2.7
BenchmarkDeepSeek V4 Pro 0813MiniMax M2.7Leader
AA Intelligence Index36.022.8DeepSeek V4 Pro 0813
Epoch Capabilities Index155.6145.8DeepSeek V4 Pro 0813
Agentic workLike-for-likePublic boards, shown as published, kept out of the index50
Agentic work benchmarks for DeepSeek V4 Pro 0813 and MiniMax M2.7
BenchmarkDeepSeek V4 Pro 0813MiniMax M2.7Leader
AA Agentic Index41.315.3DeepSeek V4 Pro 0813
SREBench (Vals AI)1.9%Not measuredOnly DeepSeek V4 Pro 0813 measured
Finance Agent (Vals AI)50.4%27.9%DeepSeek V4 Pro 0813
Tax Agent Bench (Vals AI)26.8%6.9%DeepSeek V4 Pro 0813
Legal Research Bench (Vals AI)40.9%10.6%DeepSeek V4 Pro 0813
Harvey's Legal Agent Benchmark (Vals AI)10.8%Not measuredOnly DeepSeek V4 Pro 0813 measured
Excel Modeling Benchmark (Vals AI)52.8%28.4%DeepSeek V4 Pro 0813
Terminal-Bench-Science (Vals AI)4.3%Not measuredOnly DeepSeek V4 Pro 0813 measured
τ²-Bench TelecomNot measured84.8%Only MiniMax M2.7 measured
APEX-Agents47.3%Not measuredOnly DeepSeek V4 Pro 0813 measured
Neither model has a result on the other 29 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index60
Reasoning benchmarks for DeepSeek V4 Pro 0813 and MiniMax M2.7
BenchmarkDeepSeek V4 Pro 0813MiniMax M2.7Leader
GPQA Diamond92.4%86.6%DeepSeek V4 Pro 0813
GPQA Diamond (Artificial Analysis run)92.8%87.4%DeepSeek V4 Pro 0813
Humanity's Last Exam (Artificial Analysis run)41.0%29.6%DeepSeek V4 Pro 0813
GPQA Diamond (Epoch AI run)91.7%Not measuredOnly DeepSeek V4 Pro 0813 measured
ARC-AGI90.5%Not measuredOnly DeepSeek V4 Pro 0813 measured
ARC-AGI-261.3%Not measuredOnly DeepSeek V4 Pro 0813 measured
Chess puzzles46.0%Not measuredOnly DeepSeek V4 Pro 0813 measured
Mystery game puzzles43.0%Not measuredOnly DeepSeek V4 Pro 0813 measured
LMArena Text: hard prompts14581422DeepSeek V4 Pro 0813
LMArena Text: expert14591443DeepSeek V4 Pro 0813
LMArena Text: hard prompts in English14631441DeepSeek V4 Pro 0813
Neither model has a result on the other 10 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index30
Math benchmarks for DeepSeek V4 Pro 0813 and MiniMax M2.7
BenchmarkDeepSeek V4 Pro 0813MiniMax M2.7Leader
AIMENot measured91.0%Only MiniMax M2.7 measured
OTIS Mock AIME98.6%Not measuredOnly DeepSeek V4 Pro 0813 measured
FrontierMath, tiers 1 to 364.6%Not measuredOnly DeepSeek V4 Pro 0813 measured
FrontierMath, tier 426.8%Not measuredOnly DeepSeek V4 Pro 0813 measured
ProofBench50.0%3.0%DeepSeek V4 Pro 0813
LMArena Text: math14411421DeepSeek V4 Pro 0813
LMArena Text: mathematical work14431437DeepSeek V4 Pro 0813
Neither model has a result on the other 6 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index81
Knowledge benchmarks for DeepSeek V4 Pro 0813 and MiniMax M2.7
BenchmarkDeepSeek V4 Pro 0813MiniMax M2.7Leader
MMLU Pro87.0%80.4%DeepSeek V4 Pro 0813
LegalBench (Vals AI)82.4%84.0%MiniMax M2.7
CaseLaw (Vals AI)Not measured60.9%Only MiniMax M2.7 measured
CorpFin (Vals AI)65.4%61.2%DeepSeek V4 Pro 0813
TaxEval (Vals AI)73.1%66.6%DeepSeek V4 Pro 0813
MedCode (Vals AI)42.5%34.4%DeepSeek V4 Pro 0813
SimpleQA Verified52.9%Not measuredOnly DeepSeek V4 Pro 0813 measured
LMArena Text: science14571416DeepSeek V4 Pro 0813
LMArena Text: medicine and healthcare14461409DeepSeek V4 Pro 0813
LMArena Text: legal and government14591403DeepSeek V4 Pro 0813
LMArena Text: business and finance14421407DeepSeek V4 Pro 0813
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 DeepSeek V4 Pro 0813 and MiniMax M2.7
BenchmarkDeepSeek V4 Pro 0813MiniMax M2.7Leader
Design Arena: websitesNot measured1243Only MiniMax M2.7 measured
Design Arena: UI componentsNot measured1210Only MiniMax M2.7 measured
Design Arena: game devNot measured1210Only MiniMax M2.7 measured
Design Arena: data vizNot measured1233Only MiniMax M2.7 measured
Design Arena: 3DNot measured1195Only MiniMax M2.7 measured
Design Arena: all code categoriesNot measured1233Only MiniMax M2.7 measured
Design Arena: SVGNot measured1136Only MiniMax M2.7 measured
Design Arena: ASCII artNot measured1136Only MiniMax M2.7 measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index160
General benchmarks for DeepSeek V4 Pro 0813 and MiniMax M2.7
BenchmarkDeepSeek V4 Pro 0813MiniMax M2.7Leader
Vals Index53.0%Not measuredOnly DeepSeek V4 Pro 0813 measured
MedScribe (Vals AI)80.2%79.9%DeepSeek V4 Pro 0813
IFBenchNot measured75.7%Only MiniMax M2.7 measured
Long context reasoning (AA-LCR)80.3%78.3%DeepSeek V4 Pro 0813
DTBench93.9%Not measuredOnly DeepSeek V4 Pro 0813 measured
LMCA45.5%Not measuredOnly DeepSeek V4 Pro 0813 measured
ExploitBenchNot measured13.3%Only MiniMax M2.7 measured
LMArena Text14461405DeepSeek V4 Pro 0813
LMArena Text: instruction following14481405DeepSeek V4 Pro 0813
LMArena Text: creative writing14391354DeepSeek V4 Pro 0813
LMArena Text: English14571426DeepSeek V4 Pro 0813
LMArena Text: non-English14311382DeepSeek V4 Pro 0813
LMArena Text: Chinese14811442DeepSeek V4 Pro 0813
LMArena Text: GermanNot measured1398Only MiniMax M2.7 measured
LMArena Text: JapaneseNot measured1260Only MiniMax M2.7 measured
LMArena Text: Korean14491314DeepSeek V4 Pro 0813
LMArena Text: French14461424DeepSeek V4 Pro 0813
LMArena Text: Spanish14351402DeepSeek V4 Pro 0813
LMArena Text: Russian14481383DeepSeek V4 Pro 0813
LMArena Text: PolishNot measured1387Only MiniMax M2.7 measured
LMArena Text: multi-turn14541412DeepSeek V4 Pro 0813
LMArena Text: longer queries14581418DeepSeek V4 Pro 0813
LMArena Text: writing and language14431376DeepSeek V4 Pro 0813
LMArena Text: entertainment and media14241360DeepSeek V4 Pro 0813
Neither model has a result on the other 22 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for DeepSeek V4 Pro 0813 and MiniMax M2.7
BenchmarkDeepSeek V4 Pro 0813MiniMax M2.7Leader
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 0813$0.0016Fits in one request
MiniMax M2.7$0.0009Fits in one request

MiniMax M2.7 costs less

Repository review

50K input + 3K output tokens

DeepSeek V4 Pro 0813$0.039Fits in one request
MiniMax M2.7$0.019Fits in one request

MiniMax M2.7 costs less

Agent loop

200K input + 10K output tokens

DeepSeek V4 Pro 0813$0.152Fits in one request
MiniMax M2.7$0.072Exceeds the context window

MiniMax M2.7 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 0813 and MiniMax M2.7
SpecificationDeepSeek V4 Pro 0813MiniMax M2.7
MakerDeepSeekMiniMax
WeightsOpen weightsOpen weights
ReleasedAugust 12, 2026March 18, 2026
Context window1.05M205K
Input $/Mtok$0.66$0.30
Output $/Mtok$1.98$1.20
Coding index54.628.4
Independent measurements behind it11
Cost per solved task$0.10$0.64
SourceOpenRouter listingOpenRouter listing

Questions

Is DeepSeek V4 Pro 0813 better than MiniMax M2.7 for coding?

Pick DeepSeek V4 Pro 0813 for coding: it scores 54.6 on the Vibe Coding Index against 28.4 for MiniMax M2.7. Pick MiniMax M2.7 when output tokens cost 1.7x less, or when the benchmark that matters is LegalBench (Vals AI), where it leads.

Which is cheaper, DeepSeek V4 Pro 0813 or MiniMax M2.7?

DeepSeek V4 Pro 0813 costs $0.66 per million input tokens and $1.98 per million output tokens; MiniMax M2.7 costs $0.30 and $1.20. Per solved SWE-bench Verified task, DeepSeek V4 Pro 0813 costs $0.10 and MiniMax M2.7 $0.64.

Which has the bigger context window, DeepSeek V4 Pro 0813 or MiniMax M2.7?

DeepSeek V4 Pro 0813, with 1.05M tokens against 205K.

How many benchmarks is the DeepSeek V4 Pro 0813 vs MiniMax M2.7 comparison based on?

56 benchmarks have a result for both models as of October 21, 2026. DeepSeek V4 Pro 0813's coding index rests on 1 independent benchmark and MiniMax M2.7's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next