MMODELYST
Models/LFM2 2.6B
MDL

LFM2 2.6B

Liquid AI

Verified against Artificial Analysis · Jul 27, 2026
Capability
22
percentile index
Price /M
$0 / $0
in / out
Context
Avg score
18.2
11 benchmarks
AA Intelligence
2.7
index
Weights
Closed
Released
Sep 23, 2025
About

LFM2 2.6B is a proprietary model in the LFM2 family from Liquid AI. Benchmarked on 11 evals, averaging 18.2.

Benchmark scores · 11
avg 18.2 — higher is better
MATH
75.3
GPQA Diamond
31.5
MMLU-Pro
29.8
IFBench
25.7
τ²-bench
13.5
AIME
8.3
LiveCodeBench
8.1
Humanity's Last Exam
5.2
SciCode
2.5
Terminal-Bench Hard
0.8
AA Long-Context Reasoning
0
The research behind this model
all papers mentioning it →

Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.

Details
PublisherLiquid AI
FamilyLFM2
Parameters
License
WeightsClosed
ReleasedSep 23, 2025
Capability over time
GPQA Diamond29.2 → 31.5
yesterday
MATH76.8 → 75.3
yesterday
IFBench26.5 → 25.7
yesterday
GPQA Diamond30.6 → 29.2
8d ago
MATHfirst seen at 76.8
8d ago
IFBench19.5 → 26.5
8d ago

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

API providers
Artificial Analysis$0/M