Vascend
HorizonVascend IDEVascend CLICommunityBenchmarkContact
Sign inJoin Discord
HomeHorizonIDECLISign in
ChangelogIntelligenceCost per taskCoding Agent IndexCyber IndexDecision IndexBenchmarksSpeedLatencyPriceLeaderboard
Benchmark

LFM2.5-1.2B-Instruct

  • Liquid AI
  • Release · January 5, 2026
  • Open weights
  • Non-reasoning

Intelligence

Intelligence Index

#249 of 266
  1. 5.55.5Ministral 3 8B
  2. 5.45.4Qwen3.5 0.8B (Non-reasoning)
  3. 5.45.4Llama 3.2 Instruct 11B (Vision)
  4. 5.35.3Exaone 4.0 1.2B (Reasoning)
  5. 5.25.2LFM2.5-1.2B-Thinking
  6. 5.25.2LFM2.5-1.2B-Instruct
  7. 5.25.2Jamba 1.7 Mini
  8. 5.25.2Olmo 3 7B Instruct
  9. 5.25.2Granite 4.0 H 1B
  10. 5.25.2Exaone 4.0 1.2B (Non-reasoning)
Data from Artificial Analysis

Evaluations

  • AA-Omniscience Accuracy7%#240
  • Humanity's Last Exam6.7%#178
  • AA-LCR0%#229
  • CritPt0%#143
  • GPQA Diamond32.6%#196
  • IFBench43.8%#74
  • τ²-Bench Telecom10.8%#138
  • Terminal-Bench Hard0%#121
LFM2.5-1.2B-InstructMedian
Data from Artificial Analysis

Nearby models

  • Ministral 3 8BMistral5.5
  • Qwen3.5 0.8B (Non-reasoning)Alibaba5.4
  • Llama 3.2 Instruct 11B (Vision)Meta5.4
  • Exaone 4.0 1.2B (Reasoning)LG AI Research5.3
  • LFM2.5-1.2B-ThinkingLiquid AI5.2
  • Jamba 1.7 MiniAI21 Labs5.2
  • Olmo 3 7B InstructAllen Institute for AI5.2
  • Granite 4.0 H 1BIBM5.2
Data from Artificial Analysis

Same lab

  • LFM2.5-2.6BLiquid AI8.4
  • LFM2.5-8B-A1BLiquid AI7.2
  • LFM2 24B A2BLiquid AI5.9
  • LFM2.5-VL-1.6BLiquid AI4.8
Data from Artificial Analysis
HorizonVascend IDEOlympusTalosCoursesCommunityBenchmark
DiscordGitHubXLet’s talkLegalPrivacyCookiesData useSecurity
© 2026 Vascend Technologies · VAT IT04911550616
ascend