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

Nemotron 3 Ultra 550B A55B (Reasoning)

  • NVIDIA
  • Open weights
  • Reasoning

Intelligence

Intelligence Index

#94 of 266
  1. 25.225Qwen3.7 Plus
  2. 2525Inkling (Xhigh)
  3. 24.725Solar Open2 250B
  4. 24.625Ling-3.0-flash-VL
  5. 24.124Solar Mini 4
  6. 22.923Nemotron 3 Ultra 550B A55B (Reasoning)
  7. 22.623Ling-3.0-flash-Fin
  8. 22.222Gemini 3.5 Flash-Lite
  9. 21.822G9v3-39A5B
  10. 21.722KAT Coder Pro V2
Data from Artificial Analysis

Evaluations

  • AA-Omniscience Accuracy22.6%#102
  • Humanity's Last Exam28.4%#87
  • Terminal-Bench 4.00.5%#86
  • AA-LCR79.3%#59
  • SciCode40.3%#101
  • CritPt3.1%#90
  • AA-AnalystAgent6.3%#25
  • GPQA Diamond86.7%#40
  • IFBench81.4%#2
  • τ²-Bench Telecom83.3%#31
  • τ²-Bench Banking14.2%#62
  • Terminal-Bench Hard36.4%#13
Nemotron 3 Ultra 550B A55B (Reasoning)Median
Data from Artificial Analysis

Speed and price

Speed

tokens per second · #43 of 162
  1. 168168NVIDIA Nemotron Nano 9B V2 (Non-reasoning)
  2. 166166Command A+
  3. 162162Mistral Medium 3.5
  4. 162162Muse Wonder (Max)
  5. 159159Quasar 438B (Max, Based on GLM-5.2)
  6. 158158Nemotron 3 Ultra 550B A55B (Reasoning)
  7. 152152GPT-5.5 Instant (June 2026)
  8. 152152o3
  9. 151151Muse Spark Open (Xhigh)
  10. 147147Qwen3.6 35B A3B (Non-reasoning)
Data from Artificial Analysis

Latency

seconds to first token · #48 of 162
  1. 11NVIDIA Nemotron Nano 9B V2 (Non-reasoning)
  2. 1.11.1NVIDIA Nemotron Nano 12B v2 VL (Non-reasoning)
  3. 1.11.1Muse Glimmer (High)
  4. 1.11.1Gemma 4 31B (Reasoning)
  5. 1.21.2Step 5 Preview
  6. 1.21.2Nemotron 3 Ultra 550B A55B (Reasoning)
  7. 1.21.2DeepSeek V4.1 Flash 1017 (Max)
  8. 1.31.3Quasar 438B (Max, Based on GLM-5.2)
  9. 1.31.3Trinity Large Thinking
  10. 1.31.3MiniMax-M3
Data from Artificial Analysis

Price

$ per 1M tokens (3:1) · #102 of 169
  1. $1.11.1Qwen3.5 122B A10B (Non-reasoning)
  2. $1.11.1Qwen3.5 122B A10B (Reasoning)
  3. $1.131.13Qwen3.8 27B (Xhigh)
  4. $1.131.13Qwen3.8 27B (Medium)
  5. $1.131.13Qwen3.8 27B (Low)
  6. $1.131.13Nemotron 3 Ultra 550B A55B (Reasoning)
  7. $1.131.13Qwen3.8 27B (Non-reasoning)
  8. $1.31.3LongCat 2.0
  9. $1.351.35Qwen3.5 397B A17B (Non-reasoning)
  10. $1.351.35Qwen3.5 397B A17B (Reasoning)
Data from Artificial Analysis

Nearby models

  • Qwen3.7 PlusAlibaba25.2
  • Inkling (Xhigh)Thinking Machines25
  • Solar Open2 250BUpstage24.7
  • Ling-3.0-flash-VLInclusionAI24.6
  • Solar Mini 4Upstage24.1
  • Ling-3.0-flash-FinInclusionAI22.6
  • Gemini 3.5 Flash-LiteGoogle22.2
  • G9v3-39A5BAI9Stars21.8
Data from Artificial Analysis

Same lab

  • Nemotron 3.5 Super VLNVIDIA26.5
  • Nemotron 3.5 Bolt 74B-A9BNVIDIA21.4
  • Nemotron 3.5 LightningNVIDIA12.9
  • Nemotron Cascade 2 30B A3BNVIDIA11.7
  • Nemotron 3 Nano Omni 30B A3B ReasoningNVIDIA10.3
  • NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)NVIDIA8.9
  • Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)NVIDIA7.5
  • NVIDIA Nemotron Nano 12B v2 VL (Reasoning)NVIDIA7.5
Data from Artificial Analysis
HorizonVascend IDEOlympusTalosCoursesCommunityBenchmark
DiscordGitHubXLet’s talkLegalPrivacyCookiesData useSecurity
© 2026 Vascend Technologies · VAT IT04911550616
ascend