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

Llama 3.2 Instruct 90B (Vision)

  • Meta
  • Release · September 25, 2024
  • Open weights
  • Non-reasoning

Intelligence

Intelligence Index

#223 of 266
  1. 6.66.6Sarvam 30B (High)
  2. 6.56.5Gemma 4 E2B (Non-reasoning)
  3. 6.56.5Llama 4 Scout
  4. 6.56.5Olmo 3.1 32B Instruct
  5. 6.46.4R1 1776
  6. 6.46.4Llama 3.2 Instruct 90B (Vision)
  7. 6.36.3EXAONE 4.0 32B (Non-reasoning)
  8. 6.36.3Celeris-1
  9. 6.36.3Phi-4 Mini Instruct
  10. 6.26.2Qwen3.5 2B (Non-reasoning)
Data from Artificial Analysis

Evaluations

  • Humanity's Last Exam4.5%#229
  • MMMU-Pro39.5%#112
  • GPQA Diamond43.2%#180
Llama 3.2 Instruct 90B (Vision)Median
Data from Artificial Analysis

Nearby models

  • Sarvam 30B (High)Sarvam6.6
  • Gemma 4 E2B (Non-reasoning)Google6.5
  • Llama 4 ScoutMeta6.5
  • Olmo 3.1 32B InstructAllen Institute for AI6.5
  • R1 1776Perplexity6.4
  • EXAONE 4.0 32B (Non-reasoning)LG AI Research6.3
  • Celeris-1Celeris6.3
  • Phi-4 Mini InstructMicrosoft6.3
Data from Artificial Analysis

Same lab

  • Muse Wonder Full (Max)Meta55.8
  • Muse Wonder (Max)Meta52.4
  • Muse Spark Open (Xhigh)Meta38.9
  • Muse Glimmer (High)Meta17.5
  • Llama 4 MaverickMeta9.3
  • Llama 3.3 Instruct 70BMeta7.7
  • Llama 3.1 Instruct 405BMeta7.3
  • Llama 3.2 Instruct 11B (Vision)Meta5.4
Data from Artificial Analysis
HorizonVascend IDEOlympusTalosCoursesCommunityBenchmark
DiscordGitHubXLet’s talkLegalPrivacyCookiesData useSecurity
© 2026 Vascend Technologies · VAT IT04911550616
ascend