Inference
EN
LiquidAI

LFM2.5-2.6B

LFM2.5-2.6B is a reasoning model from LiquidAI. 13 benchmarks count toward its index, in 6 categories.

  • Open weights
  • Reasoning
  • Input: text
  • Released 2026-08
  • Retires 13 Nov 2026
RankedEnough results for an index.
IndexAll categories, out of 100.29.3±6.0Median 51.0
CoverageIndex weight that has results.85%
SpeedOutput tokens per second.189/sMedian 67/s
First tokenSeconds to the first output token.0.2 sMedian 1.3 s
Input / 1MUS dollars per 1M input tokens.$0.35Median $0.60
Output / 1MUS dollars per 1M output tokens.$0.50Median $2.55
ContextMaximum tokens in one request.128KMedian 262K
EloLMArena rating and rank.—Median 1422

Details

From OpenRouter
API IDThe model ID on OpenRouter.
liquid/lfm-2.5-2.6b
Max outputMaximum output tokens per request.
8K
Knowledge cutoffLast date of training data.
—
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
$0.04
WeightsLink to the published weights.
Hugging Face (opens in a new tab) ↗
RetiresDate OpenRouter removes the model.
13 Nov 2026

Run locally

2.7B parameters
16-bit
6.7 GB
8-bit
4.0 GB
4-bit
2.7 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy fits in 8 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.5 counted32.0196 / 218
CodingCode writing and repair.2 counted23.9204 / 209
ReasoningLogic problems and puzzles.2 counted24.0219 / 227
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.2 counted29.5240 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.1 counted44.3—
MathMath problems.1 counted27.2—

Results

13 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
Claw-EvalAgenticAbove fitFar above this model's index.62.9%22 / 42
Berkeley Function Calling Leaderboard v4Agentic56.9%15 / 26
Critical Physics TasksReasoning0.0%150 / 214
GDPval-AA normalizedAgentic0.0%101 / 133
Artificial Analysis Humanity's Last ExamKnowledge6.2%169 / 222
Instruction Following BenchmarkInstruction59.2%44 / 57
American Invitational Mathematics Examination 2025Math51.9%41 / 53
LiveCodeBench v6Coding59.4%49 / 59
Artificial Analysis GPQA DiamondKnowledge55.8%167 / 199
PinchBenchAgentic68.2%9 / 10
Artificial Analysis Long Context ReasoningReasoning5.7%202 / 217
Artificial Analysis SciCodeCoding14.4%114 / 114
τ³-Bench Tool-Agent-User EvaluationAgentic5.7%32 / 32

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuse

History

5 releases
Index29.0 → 29.3
23 Oct 202612 Nov 2026
Output / 1MFree → $0.50
23 Oct 202612 Nov 2026

Questions

How much does LFM2.5-2.6B cost?
$0.35 per 1M input tokens, $0.50 per 1M output tokens. A free tier is also available.
How good is LFM2.5-2.6B?
Its index is 29.3 out of 100. That is place 252 of 283 ranked models.
What is LFM2.5-2.6B best at?
Knowledge: 29.5 out of 100, place 240 of 268.
How fast is LFM2.5-2.6B?
189 output tokens per second, 0.2 s to the first token.
What is the context window of LFM2.5-2.6B?
128K tokens.
Is LFM2.5-2.6B open source?
Yes. Its weights are published.
When was LFM2.5-2.6B released?
August 2026.

Badge

LFM2.5-2.6B on Inference360: place 252 of 283, index 29.3
HTML
<a href="https://inference360.ai/models/lfm2-5-2-6b"><img src="https://inference360.ai/badges/lfm2-5-2-6b.svg" alt="LFM2.5-2.6B on Inference360: place 252 of 283, index 29.3" height="20"></a>
Markdown
[![LFM2.5-2.6B on Inference360: place 252 of 283, index 29.3](https://inference360.ai/badges/lfm2-5-2-6b.svg)](https://inference360.ai/models/lfm2-5-2-6b)

Compare with

Gemini 2.0 Flash29.5GPT-4o28.7DeepSeek V329.9LFM2.5-8B-A1B30.6LFM2.5-24B-A2B39.8

Lower price

dots3-note Preview63.4 · FreeApodex 1.1 Mini55.7 · FreeNemotron 3 Ultra50.8 · Free

More from LiquidAI

All models
LFM2.5-VL-3BUnrankedLFM2.5-230MUnrankedLFM2.5-ColBERT-350MUnrankedd1-3BUnrankedd1-omni-600MUnranked