Enter opens the highlighted result. Esc closes.

Llama 3.3 70B Instruct vs MiMo-V2.5-Pro

MiMo-V2.5-Pro leads Llama 3.3 70B Instruct on the Artificial Analysis Intelligence Index, 26.0 to 7.7, and Llama 3.3 70B Instruct costs 1.3x less per output token ($0.72 vs $0.9152 per million). Updated October 22, 2026; 38 benchmarks have a result for both.

Replace Llama 3.3 70B Instruct Replace MiMo-V2.5-Pro
Llama 3.3 70B InstructMeta Llama
Coding
No data
Overall
26.0
Output $/M
$0.72
0
MiMo-V2.5-ProXiaomi
Coding
31.6
Overall
39.3
Output $/M
$0.9152
38
vs boards won of 38 shared

Llama 3.3 70B Instruct or MiMo-V2.5-Pro: which to pick

Only MiMo-V2.5-Pro has a coding score, 31.6 on the Vibe Coding Index; Llama 3.3 70B Instruct has none, so the two cannot be ranked on coding.

Pick Llama 3.3 70B Instruct for output tokens 1.3x cheaper ($0.72 vs $0.9152 per million).

Model ALlama 3.3 70B InstructMeta LlamaNot benchmarkedNo coding benchmarkNo cost per solved taskModel BMiMo-V2.5-ProXiaomi31.6/1004 independent benchmarks$0.03 per solved task
Shared results
38
Llama 3.3 70B Instruct only
14
MiMo-V2.5-Pro only
32
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
Llama 3.3 70B Instruct not benchmarked, MiMo-V2.5-Pro 31.6 from 4 benchmarks.
Widest category
Agentic work, 67.5 points apart on average over 1 shared benchmark.
Widest single gap
τ²-Bench Telecom: Llama 3.3 70B Instruct 26.6% against MiMo-V2.5-Pro 94.2%, 67.5 points apart on a 0 to 100 scale.
Output price
Llama 3.3 70B Instruct $0.72 against MiMo-V2.5-Pro $0.9152 per million output tokens, 1.3x apart.
Context window
Llama 3.3 70B Instruct takes 131K tokens, MiMo-V2.5-Pro 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
CategoryLlama 3.3 70B InstructBarsMiMo-V2.5-ProTies
Coding04
Intelligence01
Agentic work01
Reasoning05
Math02
Knowledge04
General021
All shared boards038
CodingLike-for-likeSix of these boards feed the Vibe Coding Index04
Coding benchmarks for Llama 3.3 70B Instruct and MiMo-V2.5-Pro
BenchmarkLlama 3.3 70B InstructMiMo-V2.5-ProLeader
LMArena WebDevNot measured1480Only MiMo-V2.5-Pro measured
AA Coding Index11.960.2MiMo-V2.5-Pro
LiveCodeBench (Vals AI)Not measured81.3%Only MiMo-V2.5-Pro measured
SWE-bench Verified (Vals AI)Not measured74.0%Only MiMo-V2.5-Pro measured
Terminal-Bench 4 (Vals AI)Not measured0.5%Only MiMo-V2.5-Pro measured
Vibe Code Bench (Vals AI)Not measured34.1%Only MiMo-V2.5-Pro measured
Code migration (Vals AI)Not measured21.6%Only MiMo-V2.5-Pro measured
Terminal-Bench 2.1 (Vals AI)Not measured57.3%Only MiMo-V2.5-Pro measured
LiveCodeBench (Artificial Analysis run)28.8%Not measuredOnly Llama 3.3 70B Instruct measured
SciCodeNot measured50.6%Only MiMo-V2.5-Pro measured
Terminal-Bench Hard (Artificial Analysis run)3.0%43.2%MiMo-V2.5-Pro
WeirdML14.4%Not measuredOnly Llama 3.3 70B Instruct measured
ALE-BenchNot measured899.8Only MiMo-V2.5-Pro measured
LMArena Text: coding12681501MiMo-V2.5-Pro
LMArena Text: software and IT12731495MiMo-V2.5-Pro
LMArena WebDev: ReactNot measured1469Only MiMo-V2.5-Pro measured
LMArena WebDev: HTMLNot measured1473Only MiMo-V2.5-Pro measured
Neither model has a result on the other 22 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
Intelligence benchmarks for Llama 3.3 70B Instruct and MiMo-V2.5-Pro
BenchmarkLlama 3.3 70B InstructMiMo-V2.5-ProLeader
AA Intelligence Index7.726.0MiMo-V2.5-Pro
Epoch Capabilities Index127.3Not measuredOnly Llama 3.3 70B Instruct measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
Agentic work benchmarks for Llama 3.3 70B Instruct and MiMo-V2.5-Pro
BenchmarkLlama 3.3 70B InstructMiMo-V2.5-ProLeader
AA Agentic IndexNot measured21.3Only MiMo-V2.5-Pro measured
Finance Agent (Vals AI)Not measured41.5%Only MiMo-V2.5-Pro measured
Tax Agent Bench (Vals AI)Not measured11.5%Only MiMo-V2.5-Pro measured
Legal Research Bench (Vals AI)Not measured15.9%Only MiMo-V2.5-Pro measured
Excel Modeling Benchmark (Vals AI)Not measured55.2%Only MiMo-V2.5-Pro measured
τ²-Bench Telecom26.6%94.2%MiMo-V2.5-Pro
BALROG23.0%Not measuredOnly Llama 3.3 70B Instruct measured
Berkeley Function Calling Leaderboard (BFCL)31.9%Not measuredOnly Llama 3.3 70B Instruct measured
Neither model has a result on the other 33 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index05
Reasoning benchmarks for Llama 3.3 70B Instruct and MiMo-V2.5-Pro
BenchmarkLlama 3.3 70B InstructMiMo-V2.5-ProLeader
GPQA DiamondNot measured82.6%Only MiMo-V2.5-Pro measured
GPQA Diamond (Artificial Analysis run)49.8%86.6%MiMo-V2.5-Pro
Humanity's Last Exam (Artificial Analysis run)3.6%35.7%MiMo-V2.5-Pro
GPQA Diamond (Epoch AI run)47.4%Not measuredOnly Llama 3.3 70B Instruct measured
SimpleBench19.9%Not measuredOnly Llama 3.3 70B Instruct measured
LMArena Text: hard prompts12571486MiMo-V2.5-Pro
LMArena Text: expert12251501MiMo-V2.5-Pro
LMArena Text: hard prompts in English12771494MiMo-V2.5-Pro
Neither model has a result on the other 14 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index02
Math benchmarks for Llama 3.3 70B Instruct and MiMo-V2.5-Pro
BenchmarkLlama 3.3 70B InstructMiMo-V2.5-ProLeader
AA Math Index7.7Not measuredOnly Llama 3.3 70B Instruct measured
AIME 2025 (Artificial Analysis run)7.7%Not measuredOnly Llama 3.3 70B Instruct measured
OTIS Mock AIME5.1%Not measuredOnly Llama 3.3 70B Instruct measured
ProofBenchNot measured22.0%Only MiMo-V2.5-Pro measured
MATH level 541.6%Not measuredOnly Llama 3.3 70B Instruct measured
LMArena Text: math12671481MiMo-V2.5-Pro
LMArena Text: mathematical work12671491MiMo-V2.5-Pro
Neither model has a result on the other 6 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index04
Knowledge benchmarks for Llama 3.3 70B Instruct and MiMo-V2.5-Pro
BenchmarkLlama 3.3 70B InstructMiMo-V2.5-ProLeader
MMLU ProNot measured84.6%Only MiMo-V2.5-Pro measured
LegalBench (Vals AI)Not measured77.1%Only MiMo-V2.5-Pro measured
CorpFin (Vals AI)Not measured61.4%Only MiMo-V2.5-Pro measured
TaxEval (Vals AI)Not measured73.8%Only MiMo-V2.5-Pro measured
MedCode (Vals AI)Not measured32.5%Only MiMo-V2.5-Pro measured
MMLU Pro (Artificial Analysis run)71.3%Not measuredOnly Llama 3.3 70B Instruct measured
MMLU86.3%Not measuredOnly Llama 3.3 70B Instruct measured
LMArena Text: science12841482MiMo-V2.5-Pro
LMArena Text: medicine and healthcare12751473MiMo-V2.5-Pro
LMArena Text: legal and government12841474MiMo-V2.5-Pro
LMArena Text: business and finance12431466MiMo-V2.5-Pro
Neither model has a result on the other 7 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
Web design benchmarks for Llama 3.3 70B Instruct and MiMo-V2.5-Pro
BenchmarkLlama 3.3 70B InstructMiMo-V2.5-ProLeader
Design Arena: websitesNot measured1266Only MiMo-V2.5-Pro measured
Design Arena: UI componentsNot measured1246Only MiMo-V2.5-Pro measured
Design Arena: game devNot measured1261Only MiMo-V2.5-Pro measured
Design Arena: data vizNot measured1257Only MiMo-V2.5-Pro measured
Design Arena: 3DNot measured1243Only MiMo-V2.5-Pro measured
Design Arena: all code categoriesNot measured1265Only MiMo-V2.5-Pro measured
Design Arena: SVGNot measured1181Only MiMo-V2.5-Pro measured
Design Arena: ASCII artNot measured1156Only MiMo-V2.5-Pro measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index021
General benchmarks for Llama 3.3 70B Instruct and MiMo-V2.5-Pro
BenchmarkLlama 3.3 70B InstructMiMo-V2.5-ProLeader
MedScribe (Vals AI)Not measured83.7%Only MiMo-V2.5-Pro measured
IFBench47.1%79.9%MiMo-V2.5-Pro
Long context reasoning (AA-LCR)15.7%79.7%MiMo-V2.5-Pro
DTBench59.5%84.5%MiMo-V2.5-Pro
LMCA17.5%29.5%MiMo-V2.5-Pro
Fiction.LiveBench33.3%Not measuredOnly Llama 3.3 70B Instruct measured
LMArena Text12741464MiMo-V2.5-Pro
LMArena Text: instruction following12421475MiMo-V2.5-Pro
LMArena Text: creative writing12501438MiMo-V2.5-Pro
LMArena Text: English13031477MiMo-V2.5-Pro
LMArena Text: non-English12361448MiMo-V2.5-Pro
LMArena Text: Chinese12171507MiMo-V2.5-Pro
LMArena Text: German12511459MiMo-V2.5-Pro
LMArena Text: Japanese11501410MiMo-V2.5-Pro
LMArena Text: Korean11421435MiMo-V2.5-Pro
LMArena Text: French12821487MiMo-V2.5-Pro
LMArena Text: Spanish12691469MiMo-V2.5-Pro
LMArena Text: Russian12521449MiMo-V2.5-Pro
LMArena Text: Polish12591464MiMo-V2.5-Pro
LMArena Text: multi-turn12801476MiMo-V2.5-Pro
LMArena Text: longer queries12551481MiMo-V2.5-Pro
LMArena Text: writing and language12551451MiMo-V2.5-Pro
LMArena Text: entertainment and media12411438MiMo-V2.5-Pro
Neither model has a result on the other 23 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Llama 3.3 70B Instruct and MiMo-V2.5-Pro
BenchmarkLlama 3.3 70B InstructMiMo-V2.5-ProLeader
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

Llama 3.3 70B Instruct$0.0008Fits in one request
MiMo-V2.5-Pro$0.0009Fits in one request

Llama 3.3 70B Instruct costs less

Repository review

50K input + 3K output tokens

Llama 3.3 70B Instruct$0.025Fits in one request
MiMo-V2.5-Pro$0.026Fits in one request

Llama 3.3 70B Instruct costs less

Agent loop

200K input + 10K output tokens

Llama 3.3 70B Instruct$0.097Exceeds the context window
MiMo-V2.5-Pro$0.101Fits in one request

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 Llama 3.3 70B Instruct and MiMo-V2.5-Pro
SpecificationMeta Llama 3.3 70B InstructXiaomi MiMo-V2.5-Pro
MakerMeta LlamaXiaomi
WeightsOpen weightsOpen weights
ReleasedDecember 6, 2024April 22, 2026
Context window131K1.05M
Input $/Mtok$0.45$0.4575
Output $/Mtok$0.72$0.9152
Coding indexNot benchmarked31.6
Independent measurements behind it04
Cost per solved taskNot computable$0.03
SourceOpenRouter listingOpenRouter listing

Questions

Is Llama 3.3 70B Instruct better than MiMo-V2.5-Pro for coding?

Only MiMo-V2.5-Pro has a coding score, 31.6 on the Vibe Coding Index; Llama 3.3 70B Instruct has none, so the two cannot be ranked on coding. Pick Llama 3.3 70B Instruct for output tokens 1.3x cheaper ($0.72 vs $0.9152 per million).

Which is cheaper, Llama 3.3 70B Instruct or MiMo-V2.5-Pro?

Llama 3.3 70B Instruct costs $0.45 per million input tokens and $0.72 per million output tokens; MiMo-V2.5-Pro costs $0.4575 and $0.9152. Cost per solved task is known for MiMo-V2.5-Pro only.

Which has the bigger context window, Llama 3.3 70B Instruct or MiMo-V2.5-Pro?

MiMo-V2.5-Pro, with 1.05M tokens against 131K.

How many benchmarks is the Llama 3.3 70B Instruct vs MiMo-V2.5-Pro comparison based on?

38 benchmarks have a result for both models as of October 22, 2026. Llama 3.3 70B Instruct's coding index rests on 0 independent benchmarks and MiMo-V2.5-Pro's on 4. Where only one model has a result, the page says so instead of naming a winner.

Compare next