VerdictPal
Join

Models / Liquid

LFM2.5-8B-A1B

LFM · Released 2026-06-08

Liquid's compact LFM 2.5 model evaluated by Artificial Analysis on 8 Jun. Keep it in the efficiency lane until local benchmark and pricing rows are imported.

Tiny efficiency model for routing, extraction, and on-device-adjacent experiments where footprint matters more than peak reasoning.

#49 of 51 on AA Index · current snapshot
8 AA Index · 10 benchmark rows · 1 sourcesPosition is benchmark-specific — not a cross-family or cross-source ranking.
Composite20
Context128K
Input / 1M—
Output / 1M—
Knowledge cutoffopen
Output speed343 t/s
TTFT (default API)1.79s
Statuslightweight
ReasoningCodingAgenticLong context
Reasoning
56
Coding
8
Agentic
16
Long context
0

20average across 4 tested families

Benchmark placements

Where LFM2.5-8B-A1B places on each public benchmark source that publishes a row. Rank counts every model with a latest row in the same test — not a universal quality score.

BenchmarkFamilyRankScoreSourceDate
AA output speedPerformance#3/ 69Artificial Analysis2026-07-21
AA time to first tokenPerformance#30/ 69Artificial Analysis2026-07-21
IFBenchReasoning#45/ 5756Artificial Analysis2026-07-21
τ³-Bench BankingAgentic#45/ 6416Artificial Analysis2026-07-21
Humanity's Last ExamReasoning#58/ 657Artificial Analysis2026-07-21
AA-LCRLong context#64/ 640Artificial Analysis2026-07-21
Artificial Analysis Intelligence IndexReasoning#64/ 678Artificial Analysis2026-07-21
SciCodeCoding#64/ 658Artificial Analysis2026-07-21
GPQA DiamondReasoning#67/ 6951Artificial Analysis2026-07-21
Terminal-BenchAgentic#67/ 695Artificial Analysis2026-07-21
Family context

The three highest-scoring carded models in each capability family. Where LFM2.5-8B-A1B shows up, it's highlighted.

ReasoningLFM2.5-8B-A1B · 56
  1. 1
    Claude Mythos Preview
    95
  2. 2
    GPT-5.6 Sol
    94
  3. 3
    Kimi K3
    94
CodingLFM2.5-8B-A1B · 8
  1. 1
    GPT 5.5
    96
  2. 2
    Claude Opus 4.8
    96
  3. 3
    Claude Opus 4.6
    96
AgenticLFM2.5-8B-A1B · 16
  1. 1
    GPT 5.5
    100
  2. 2
    Claude Opus 4.8
    100
  3. 3
    Claude Fable 5
    100
Long context
  1. 1
    Kimi K3
    75
  2. 2
    GPT-5.1
    75
  3. 3
    GPT-5.6 Sol
    74
Newest receipts

Benchmark rows added to the public ledger for LFM2.5-8B-A1B in the last 120 days. Older rows live in the full table below.

10 recent
BenchmarkFamilyScoreSourceDays ago
AA-LCRLong context0%Artificial Analysis2
AA output speedPerformance343 t/sArtificial Analysis2
AA time to first tokenPerformance1.79sArtificial Analysis2
Artificial Analysis Intelligence IndexReasoning8.3%Artificial Analysis2
GPQA DiamondReasoning51.3%Artificial Analysis2
Humanity's Last ExamReasoning6.9%Artificial Analysis2
IFBenchReasoning55.65%Artificial Analysis2
SciCodeCoding7.8%Artificial Analysis2
τ³-Bench BankingAgentic16.08%Artificial Analysis2
Terminal-BenchAgentic4.55%Artificial Analysis2
Family coverage

A high composite that hides a weak family is a trap. These bars surface the families where this model hasn't been publicly tested, and where it leads.

Reasoning56
Coding8
Agentic16
Price vs. performance
0255075100$0.10$0.30$1$3$10$30input USD / 1M tokens · log scale →← intelligence

Hover or tab any dot for name, score, and input price.

AnthropicOpenAIMistralAlibabaDeepSeekGoogleMoonshotMiniMaxxAIMetaZhipuXiaomiMeituanCohere

Full benchmark ledger

Every catalog benchmark for LFM2.5-8B-A1B. Scores link to the original source; gaps mean no public row exists yet.

10 sourced rows

10 of 32 catalog benchmarks have a sourced row for LFM2.5-8B-A1B.

Reasoning

Coding

Agentic

Long context

Performance

Sources