LFM2.5 8B A1B
May 2026
LFM2.5 8B A1BCompare inon
Benchmarks of LFM2.5 8B A1B10
BenchmarkByScoreRank
Overall1
Knowledge4
Science2
Coding1
Legacy · saturated or no longer reported2
No benchmark matches
—per task?
self-host onlyprice
—output speed?
128Kcontext
Rankingin thinking mode?
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT-6 Lunalow×3—$0.005Function calling only with none · Chat Completions+3IOpenK2 Horizon 7Bthinking×2.9—$0.30 / $1.20Memory (4-bit): 4.2 GB → 4.5 GBBetterGPT-5.6 Lunalow×2.9—$0.01
BetterMistral Small 4thinkingOpen+56%—$0.015Memory (4-bit): 4.2 GB → 72 GB
BetterMiMo v2.5thinkingOpen×3—$0.019Memory (4-bit): 4.2 GB → 186 GBModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT-6 Lunalow×3—$0.005Function calling only with none · Chat Completions+3GPT-5.6 Lunalow×2.9—$0.01Mistral Small 4thinkingOpen+56%—$0.015Memory (4-bit): 4.2 GB → 72 GBMiMo v2.5thinkingOpen×3—$0.019Memory (4-bit): 4.2 GB → 186 GB
DeepSeek V3 0324non-thinkingOpen+35%—$0.021Memory (4-bit): 4.2 GB → 411 GBModelScoreCostPer taskWhat changes
IK2 Horizon 7BthinkingOpen×2.9—$0.30 / $1.20Memory (4-bit): 4.2 GB → 4.5 GBIK2 Horizon 3.7BthinkingOpen×2.2—$0.30 / $1.20
Gemma 4 12BthinkingOpen+97%—$0.10 / $0.30Memory (4-bit): 4.2 GB → 6 GBSApriel v1.5 15B ThinkerOpen+92%——Memory (4-bit): 4.2 GB → 7.4 GBSApriel v1.6 15B ThinkerOpen+86%——Memory (4-bit): 4.2 GB → 7.4 GBModelScoreCostPer taskWhat changes
LFM2 24B A2Bnon-thinkingOpen-18%—$0.03 / $0.12Context: 128K → 33K+1LFM2.5 1.2B ThinkingthinkingOpen-28%——Context: 128K → 33KLFM2 2.6BOpen-28%——Context: 128K → 33KLFM2.5 1.2B Instructnon-thinkingOpen-28%——Context: 128K → 33KLFM2 8B A1BOpen-32%——Context: 128K → 33KLFM2.5 VL 1.6Bnon-thinkingOpen-33%——Context: 128K → 33KLFM2 1.2BOpen-33%——Context: 128K → 33KAll Liquid AI models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · compressed✓ 16 GB · 8-bit✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
Licencelfm1.0commercial use with conditions
History
2 versionsBenchmarks
all categoriesChat & Knowledge
Humanity's Last Exam6.9#227
AA Omniscience Index-33.2#157
CritPt (physics)0#180
AA Intelligence Index 4.3.2≈7.2#274
AA Omniscience · Accuracy9.3#284
GPQA Diamond51.3#259
Coding
Agents & Tool Use
τ²-bench · AA16.1#207
Writing & Language
IFBench (instructions)55.6#101
FAQ
How much does LFM2.5 8B A1B cost?›
There is no per-unit API price: self-host only.
Is LFM2.5 8B A1B free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is LFM2.5 8B A1B, and what is it best at?›
#225 of 334 in Chat & Knowledge, #161 of 239 in Coding, #204 of 242 in Agents & Tool Use, #101 of 245 in Writing & Language.
Is LFM2.5 8B A1B better than LFM2 8B A1B?›
On AA Intelligence Index 4.3.2, LFM2.5 8B A1B scores 7.2 vs 4.9 — higher.
What is the context window of LFM2.5 8B A1B?›
128K tokens.
When was LFM2.5 8B A1B released? Is there a newer version?›
Released 2026-05-28. It is the current version in its line.
Is LFM2.5 8B A1B open source?›
Yes, the weights are open (lfm1.0 licence).
Can I run LFM2.5 8B A1B locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can LFM2.5 8B A1B read images?›
No, it accepts text only.
What is LFM2.5 8B A1B?›
LFM2.5 8B A1B is an open-weights language model from Liquid AI, released May 2026, the successor to LFM2 8B A1B.
What are the best alternatives to LFM2.5 8B A1B?›
GPT-6 Luna (better), K2 Horizon 7B (open weights).