LFM2 24B A2B

February 25, 2026Last synced Apr 7, 2026, 4:00 PM

Blended Price

$0.052/M

Input Price

$0.030/M

Output Price

$0.12/M

Speed

68 tok/s

TTFT

0.25s

Benchmark Scores

Intelligence Index

External Benchmarks

AA-Omniscience Accuracy

How LFM2 24B A2B Compares

Axes

X Axis

Y Axis

Bubble Size

Filters

Blended Price (USD)$0.00 – $30.0

AgMoBench Overall≥ 3.5

Providers

ai21-labsalibabaanthropicawsazurebaiducoheredeepseek

Show quadrants

anthropic

openai

google

meta

mistral

nvidia

kimi

xai

azure

deepseek

aws

cohere

baidu

ai21-labs

zai

reka-ai

xiaomi

alibaba

minimax

ibm

Bubble size = Context Window

16384.00

2000000.00

Compare with other models

GPQA Diamond

0.5 / 100

LiveCodeBench

— / 100

Terminal-Bench Hard

0.0 / 100

τ²-Bench

0.1 / 100

AA-Omniscience Hallucination RatePredicted

97.1 / 100

Aider PolyglotPredicted

29.9 / 100

AIMEPredicted

0.0 / 30

AIME 2025Predicted

0.0 / 30

AlpacaEval 2.0 LCPredicted

18.9 / 100

AlpacaEval 2.0 RawPredicted

15.2 / 100

ARC-AGI-1Predicted

62.5 / 100

ARC-AGI-1 Cost per TaskPredicted

0.5

ARC-AGI-2Predicted

8.8 / 100

ARC-AGI-2 Cost per TaskPredicted

0.7

BFCL (Berkeley Function Calling)Predicted

28.3

BigCodeBench CompletePredicted

29.8 / 100

BigCodeBench InstructPredicted

23.3 / 100

AA Intelligence Index (Matrix)Predicted

52.5

AA Long Context Reasoning (Matrix)Predicted

76.5

AIME 2024Predicted

59.0

AIME 2025 (Matrix)Predicted

92.7

Arena-Hard AutoPredicted

26.1

BrowseCompPredicted

85.1

BRUMO 2025Predicted

99.5

CMIMC 2025Predicted

93.3

CritPtPredicted

10.2

GPQA Diamond (Matrix)Predicted

58.1

HLE (Matrix)Predicted

33.6

HMMT Feb 2025Predicted

77.9

HMMT Nov 2025Predicted

93.9

HumanEvalPredicted

73.8

IFBench (Matrix)Predicted

42.4

IFEvalPredicted

79.7

IMO 2025Predicted

43.1

LiveCodeBench (Matrix)Predicted

45.2

MATH-500 (Matrix)Predicted

90.5

MathArena Apex 2025Predicted

13.4

MMLU-Pro (Matrix)Predicted

48.6

MMMU-ProPredicted

76.8

MRCR v2Predicted

81.3

OSWorldPredicted

75.1

SimpleQAPredicted

50.2

SMT 2025Predicted

92.9

SWE-bench ProPredicted

55.2

Tau-Bench Telecom (Matrix)Predicted

98.8

Terminal-Bench 2.0Predicted

70.7

Terminal-Bench 1.0Predicted

32.6

USAMO 2025Predicted

13.0

Video-MMUPredicted

84.4

browsecompPredicted

86.6

BullshitBenchPredicted

42.5 / 100

Aider PolyglotPredicted

0.0

Apex AgentsPredicted

4.1

Arc Agi 2Predicted

0.1

BIG-Bench HardPredicted

3.0

CAD-EvalPredicted

6.5

Chess PuzzlesPredicted

0.3

CyBenchPredicted

0.2

DeepResearchBenchPredicted

0.4

FictionLiveBenchPredicted

0.5

GeoBenchPredicted

0.0

GSM8K (Epoch)Predicted

4.8

HellaSwagPredicted

0.0

Lech Mazur WritingPredicted

7.3

METR Time HorizonsPredicted

10.3

OTIS Mock AIME 2024–2025Predicted

0.1

PosttrainbenchPredicted

0.0

SimpleQA Verified (Epoch)Predicted

0.5

The Agent CompanyPredicted

1.1

TriviaQAPredicted

10.9

WinoGrandePredicted

0.7

FrontierMathPredicted

38.8 / 100

GAIA Level 1Predicted

14.2

GAIA Level 2Predicted

1.6

GAIA Level 3Predicted

0.0

GAIAPredicted

8.4 / 100

LegalBenchPredicted

32.9 / 100

LiveBench CodingPredicted

77.3 / 100

LiveBench Data AnalysisPredicted

67.9 / 100

LiveBench LanguagePredicted

79.3 / 100

LiveBench MathPredicted

85.3 / 100

LiveBench OverallPredicted

72.3 / 100

LiveBench ReasoningPredicted

76.4 / 100

LiveCodeBenchPredicted

0.2 / 100

LongBench v2 EasyPredicted

33.3

LongBench v2 HardPredicted

27.5

LongBench v2Predicted

25.6 / 100

MATH-500Predicted

0.4 / 100

MathVistaPredicted

19.7 / 100

MLE-benchPredicted

59.7 / 100

MMLU ProPredicted

0.5 / 100

MMMUPredicted

63.7 / 100

MMTU Table UnderstandingPredicted

52.9 / 100

MT-BenchPredicted

7.7 / 10

NoLiMa (NIAH)Predicted

84.7 / 100

OCRBench v2Predicted

61.1 / 100

Open LLM AveragePredicted

17.5 / 100

Open LLM: BBHPredicted

42.2 / 100

Open LLM: GPQAPredicted

27.3 / 100

Open LLM: IFEvalPredicted

53.5 / 100

Open LLM: MATH Level 5Predicted

7.5 / 100

Open LLM: MMLU-PROPredicted

26.3 / 100

Open LLM: MUSRPredicted

37.6 / 100

RE-BenchPredicted

99.9

SimpleBenchPredicted

45.3 / 100

simpleqaPredicted

39.8

SWE-bench LitePredicted

13.9 / 100

SWE-bench VerifiedPredicted

53.3 / 100

tau-bench RetailPredicted

91.2 / 100

Vectara Factual ConsistencyPredicted

87.7 / 100

Vectara Hallucination RatePredicted

12.3 / 100

WebArenaPredicted

0.1 / 100

WeirdMLPredicted

12.7 / 100

WildBenchPredicted

30.1

HuggingFace Downloads (30d)hf-downloads

449061.0

HuggingFace Likeshf-downloads

1.0