MMODELYST
Models/LFM2.5-8B-A1B
MDL

LFM2.5-8B-A1B

Liquid AI

Verified against Artificial Analysis · Jul 27, 2026
Capability
42
percentile index
Price /M
$0 / $0
in / out
Context
Avg score
28.2
8 benchmarks
AA Intelligence
8.3
index
Speed
332 tok/s
TTFT 1.5s
Released
May 28, 2026
About

LFM2.5-8B-A1B is a proprietary model in the LFM2.5 family from Liquid AI. Benchmarked on 8 evals, averaging 28.2.

Benchmark scores · 8
avg 28.2 — higher is better
MATH
90.2
IFBench
53.3
GPQA Diamond
46.6
τ²-bench
16.1
SciCode
7.8
Humanity's Last Exam
6.9
Terminal-Bench Hard
4.5
AA Long-Context Reasoning
0
Details
PublisherLiquid AI
FamilyLFM2.5
Parameters
License
WeightsClosed
ReleasedMay 28, 2026
Performance & price
Output speed332 tok/s
Latency (TTFT)1.5s
First answer7.5s
Price in / out$0 / $0 per 1M

Median across API providers, via Artificial Analysis. Blended = 3:1 input:output. For reasoning models, latency includes thinking time. Methodology →

Capability over time
IFBench55.6 → 53.3
yesterday
MATHfirst seen at 90.2
yesterday
GPQA Diamond51.3 → 46.6
yesterday
τ²-benchfirst seen at 16.1
1mo ago
GPQA Diamondfirst seen at 51.3
1mo ago
Humanity's Last Examfirst seen at 6.9
1mo ago

Append-only ledger — every observed change to this model's numbers.

API providers
Artificial Analysis$0/M