LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

Grok 4.6 vs MiMo-V2.5

Updated October 21, 2026. 74 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Grok 4.6 Replace MiMo-V2.5
Grok 4.6xAI
Coding
76.4
Overall
44.3
Output $/M
$6.00
60
MiMo-V2.5Xiaomi
Coding
34.0
Overall
21.8
Output $/M
$0.238
14
vs boards won of 74 shared

Decision reading

Pick Grok 4.6 for coding: it scores 76.4 on the Vibe Coding Index against 34.0 for MiMo-V2.5.

Pick MiMo-V2.5 when output tokens cost 25.2x less, or when a solved task costs $0.01 against $0.82.

Model AGrok 4.6xAI76.4/1002 independent benchmarks$0.82 per solved taskModel BMiMo-V2.5Xiaomi34.0/1001 independent benchmark$0.01 per solved task
Shared results
74
Grok 4.6 only
49
MiMo-V2.5 only
6
Like-for-like categories
8 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
Grok 4.6 76.4 from 2 benchmarks, MiMo-V2.5 34.0 from 1 benchmark.
Widest category
Agentic work, 25.2 points apart on average over 5 shared benchmarks.
Widest single gap
Legal Research Bench (Vals AI): Grok 4.6 48.1% against MiMo-V2.5 9.1%, 38.9 points apart on a 0 to 100 scale.
Output price
Grok 4.6 $6.00 against MiMo-V2.5 $0.238 per million output tokens, 25.2x apart.
Context window
Grok 4.6 takes 500K tokens, MiMo-V2.5 1.05M.

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
CategoryGrok 4.6BarsMiMo-V2.5Ties
Coding112
Intelligence10
Agentic work50
Reasoning62
Math21
Knowledge64
Web design80
General215
All shared boards6014
CodingLike-for-likeFour of these boards feed the Vibe Coding Index112
Coding benchmarks for Grok 4.6 and MiMo-V2.5
BenchmarkGrok 4.6MiMo-V2.5Leader
LiveBench coding89.1%Not measuredOnly Grok 4.6 measured
LMArena WebDev16171438Grok 4.6
AA Coding Index76.856.8Grok 4.6
LiveCodeBench (Vals AI)88.2%81.5%Grok 4.6
SWE-bench Verified (Vals AI)95.6%71.0%Grok 4.6
Terminal-Bench 4 (Vals AI)17.2%Not measuredOnly Grok 4.6 measured
Vibe Code Bench (Vals AI)76.2%42.2%Grok 4.6
Code migration (Vals AI)44.6%14.2%Grok 4.6
ProgramBench (Vals AI)1.0%Not measuredOnly Grok 4.6 measured
IOI (Vals AI)47.6%Not measuredOnly Grok 4.6 measured
Terminal-Bench 2.1 (Vals AI)78.3%60.7%Grok 4.6
Vibe Code Bench 1-100 (Vals AI)14.8%Not measuredOnly Grok 4.6 measured
SkillsBench (Vals AI)55.8%Not measuredOnly Grok 4.6 measured
CyberBench (Vals AI)67.7%Not measuredOnly Grok 4.6 measured
SciCode56.5%43.9%Grok 4.6
Terminal-Bench Hard (Artificial Analysis run)Not measured41.7%Only MiMo-V2.5 measured
FrontierCode48.0%Not measuredOnly Grok 4.6 measured
FrontierSWE25.3%Not measuredOnly Grok 4.6 measured
WeirdML67.3%Not measuredOnly Grok 4.6 measured
ALE-Bench1508.1514.0Grok 4.6
LMArena Text: coding14661469MiMo-V2.5
LMArena Text: software and IT14571463MiMo-V2.5
LMArena WebDev: React16311425Grok 4.6
LMArena WebDev: HTML15971429Grok 4.6
LMArena WebDev: from an image1603Not measuredOnly Grok 4.6 measured
Neither model has a result on the other 13 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
Intelligence benchmarks for Grok 4.6 and MiMo-V2.5
BenchmarkGrok 4.6MiMo-V2.5Leader
AA Intelligence Index44.321.8Grok 4.6
Epoch Capabilities Index156.6Not measuredOnly Grok 4.6 measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index50
Agentic work benchmarks for Grok 4.6 and MiMo-V2.5
BenchmarkGrok 4.6MiMo-V2.5Leader
AA Agentic Index53.015.8Grok 4.6
Design Arena: agents building web apps1247Not measuredOnly Grok 4.6 measured
Design Arena: agents building full-stack apps1264Not measuredOnly Grok 4.6 measured
Design Arena: agents building mobile apps1234Not measuredOnly Grok 4.6 measured
Design Arena: agents building Android apps1297Not measuredOnly Grok 4.6 measured
Design Arena: agents building Godot games1276Not measuredOnly Grok 4.6 measured
Design Arena: agents building games1208Not measuredOnly Grok 4.6 measured
Design Arena: HTML slides1249Not measuredOnly Grok 4.6 measured
Design Arena: PowerPoint slides1241Not measuredOnly Grok 4.6 measured
LiveBench research84.5%Not measuredOnly Grok 4.6 measured
LiveBench browser64.0%Not measuredOnly Grok 4.6 measured
LiveBench long agentic40.7%Not measuredOnly Grok 4.6 measured
Finance Agent (Vals AI)53.7%36.7%Grok 4.6
Tax Agent Bench (Vals AI)33.1%7.8%Grok 4.6
Legal Research Bench (Vals AI)48.1%9.1%Grok 4.6
Excel Modeling Benchmark (Vals AI)62.7%55.1%Grok 4.6
BioMysteryBench (Vals AI)72.2%Not measuredOnly Grok 4.6 measured
Terminal-Bench-Science (Vals AI)11.4%Not measuredOnly Grok 4.6 measured
Vals Science Index38.1%Not measuredOnly Grok 4.6 measured
Time Horizon Index (Vals AI)5.8%Not measuredOnly Grok 4.6 measured
τ²-Bench TelecomNot measured90.6%Only MiMo-V2.5 measured
APEX-Agents65.3%Not measuredOnly Grok 4.6 measured
Vending-Bench 2$9,047Not measuredOnly Grok 4.6 measured
Neither model has a result on the other 16 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index62
Reasoning benchmarks for Grok 4.6 and MiMo-V2.5
BenchmarkGrok 4.6MiMo-V2.5Leader
GPQA Diamond94.7%81.6%Grok 4.6
LiveBench reasoning91.2%Not measuredOnly Grok 4.6 measured
MysteryMechanism (Vals AI)30.6%Not measuredOnly Grok 4.6 measured
MMMU Pro (Vals AI)Not measured80.0%Only MiMo-V2.5 measured
GPQA Diamond (Artificial Analysis run)94.9%84.9%Grok 4.6
Humanity's Last Exam (Artificial Analysis run)44.1%27.2%Grok 4.6
GPQA Diamond (Epoch AI run)94.0%Not measuredOnly Grok 4.6 measured
ARC-AGI87.5%Not measuredOnly Grok 4.6 measured
ARC-AGI-267.1%Not measuredOnly Grok 4.6 measured
SimpleBench75.9%Not measuredOnly Grok 4.6 measured
Chess puzzles33.0%Not measuredOnly Grok 4.6 measured
Mystery game puzzles34.0%Not measuredOnly Grok 4.6 measured
EBR-bench30.5%Not measuredOnly Grok 4.6 measured
LMArena Text: hard prompts14481450MiMo-V2.5
LMArena Text: expert14691459Grok 4.6
LMArena Text: hard prompts in English14451461MiMo-V2.5
LMArena Vision: diagrams12851269Grok 4.6
LMArena Vision: homework12831262Grok 4.6
Neither model has a result on the other 3 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index21
Math benchmarks for Grok 4.6 and MiMo-V2.5
BenchmarkGrok 4.6MiMo-V2.5Leader
OTIS Mock AIME99.2%Not measuredOnly Grok 4.6 measured
FrontierMath, tiers 1 to 366.0%Not measuredOnly Grok 4.6 measured
FrontierMath, tier 431.7%Not measuredOnly Grok 4.6 measured
ProofBench51.0%16.0%Grok 4.6
LMArena Text: math14171436MiMo-V2.5
LMArena Text: mathematical work14641451Grok 4.6
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index64
Knowledge benchmarks for Grok 4.6 and MiMo-V2.5
BenchmarkGrok 4.6MiMo-V2.5Leader
MMLU Pro89.4%82.9%Grok 4.6
LegalBench (Vals AI)86.3%78.9%Grok 4.6
CorpFin (Vals AI)66.2%59.9%Grok 4.6
TaxEval (Vals AI)71.1%71.8%MiMo-V2.5
MortgageTax (Vals AI)64.2%59.3%Grok 4.6
MedCode (Vals AI)44.7%31.9%Grok 4.6
SimpleQA Verified49.3%Not measuredOnly Grok 4.6 measured
LMArena Text: science14371440MiMo-V2.5
LMArena Text: medicine and healthcare14201432MiMo-V2.5
LMArena Text: legal and government14291425Grok 4.6
LMArena Text: business and finance14161432MiMo-V2.5
Neither model has a result on the other 6 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index80
Web design benchmarks for Grok 4.6 and MiMo-V2.5
BenchmarkGrok 4.6MiMo-V2.5Leader
Design Arena: websites12891266Grok 4.6
Design Arena: UI components12941257Grok 4.6
Design Arena: game dev12961249Grok 4.6
Design Arena: data viz12831249Grok 4.6
Design Arena: 3D12771216Grok 4.6
Design Arena: all code categories12921260Grok 4.6
Design Arena: SVG12391178Grok 4.6
Design Arena: ASCII art12781141Grok 4.6
LiveBench design79.8%Not measuredOnly Grok 4.6 measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index215
General benchmarks for Grok 4.6 and MiMo-V2.5
BenchmarkGrok 4.6MiMo-V2.5Leader
LiveBench data analysis75.3%Not measuredOnly Grok 4.6 measured
Public Benefits Bench (Vals AI)66.8%Not measuredOnly Grok 4.6 measured
SAGE (Vals AI)28.9%43.3%MiMo-V2.5
MedScribe (Vals AI)86.5%72.2%Grok 4.6
Vals Multimodal IndexNot measured52.8%Only MiMo-V2.5 measured
IFBenchNot measured67.1%Only MiMo-V2.5 measured
Long context reasoning (AA-LCR)81.0%73.0%Grok 4.6
DTBench97.3%Not measuredOnly Grok 4.6 measured
LMCA48.5%Not measuredOnly Grok 4.6 measured
Furniture assembly40.0%Not measuredOnly Grok 4.6 measured
LMArena Text14291428Grok 4.6
LMArena Text: instruction following14311435MiMo-V2.5
LMArena Text: creative writing14261393Grok 4.6
LMArena Vision12631247Grok 4.6
LMArena Text: English14331447MiMo-V2.5
LMArena Text: non-English14201405Grok 4.6
LMArena Text: Chinese14761469Grok 4.6
LMArena Text: German14261421Grok 4.6
LMArena Text: Japanese13781306Grok 4.6
LMArena Text: Korean13911361Grok 4.6
LMArena Text: French14551448Grok 4.6
LMArena Text: Spanish14141414Grok 4.6
LMArena Text: Russian14211396Grok 4.6
LMArena Text: Polish14301413Grok 4.6
LMArena Text: multi-turn14251445MiMo-V2.5
LMArena Text: longer queries14561445Grok 4.6
LMArena Text: writing and language14291409Grok 4.6
LMArena Text: entertainment and media14211399Grok 4.6
LMArena Vision: English12701259Grok 4.6
LMArena Vision: Chinese12891326MiMo-V2.5
LMArena Vision: reading text12721256Grok 4.6
LMArena Vision: humor12701226Grok 4.6
LMArena Vision: recognizing thingsNot measured1189Only MiMo-V2.5 measured
LMArena Vision: creative writing12621230Grok 4.6
LMArena Document1452Not measuredOnly Grok 4.6 measured
Neither model has a result on the other 11 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Grok 4.6 and MiMo-V2.5
BenchmarkGrok 4.6MiMo-V2.5Leader
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

Grok 4.6$0.0050Fits in one request
MiMo-V2.5$0.0002Fits in one request

MiMo-V2.5 costs less

Repository review

50K input + 3K output tokens

Grok 4.6$0.118Fits in one request
MiMo-V2.5$0.0067Fits in one request

MiMo-V2.5 costs less

Agent loop

200K input + 10K output tokens

Grok 4.6$0.460Fits in one request
MiMo-V2.5$0.026Fits in one request

MiMo-V2.5 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 Grok 4.6 and MiMo-V2.5
SpecificationSpaceXAI Grok 4.6Xiaomi MiMo-V2.5
MakerxAIXiaomi
WeightsNo public weights listedOpen weights
ReleasedAugust 12, 2026April 22, 2026
Context window500K1.05M
Input $/Mtok$2.00$0.119
Output $/Mtok$6.00$0.238
Coding index76.434.0
Independent measurements behind it21
Cost per solved task$0.82$0.01
SourceOpenRouter listingOpenRouter listing

Questions

Is Grok 4.6 better than MiMo-V2.5 for coding?

Pick Grok 4.6 for coding: it scores 76.4 on the Vibe Coding Index against 34.0 for MiMo-V2.5. Pick MiMo-V2.5 when output tokens cost 25.2x less, or when a solved task costs $0.01 against $0.82.

Which is cheaper, Grok 4.6 or MiMo-V2.5?

Grok 4.6 costs $2.00 per million input tokens and $6.00 per million output tokens; MiMo-V2.5 costs $0.119 and $0.238. Per solved SWE-bench Verified task, Grok 4.6 costs $0.82 and MiMo-V2.5 $0.01.

Which has the bigger context window, Grok 4.6 or MiMo-V2.5?

MiMo-V2.5, with 1.05M tokens against 500K.

How many benchmarks is the Grok 4.6 vs MiMo-V2.5 comparison based on?

74 benchmarks have a result for both models as of October 21, 2026. Grok 4.6's coding index rests on 2 independent benchmarks and MiMo-V2.5's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next