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

K2 Horizon 375B A23B

  • Institute of Foundation Models
  • Release · September 3, 2026
  • Open weights
  • Reasoning

Intelligence

Intelligence Index

#68 of 266
  1. 33.734Qwen3.8 27B (Xhigh)
  2. 33.634Motif 3
  3. 33.534Gemini 3.8 Flash (Low)
  4. 32.933GPT-6 Luna (High)
  5. 32.533GPT-5.3 Codex (Xhigh)
  6. 30.531K2 Horizon 375B A23B
  7. 30.330Beam 1.0
  8. 30.130GPT-5.6 Terra (Medium)
  9. 29.930GPT-6 Luna (Medium)
  10. 29.730Gemini 3.1 Pro Preview
Data from Artificial Analysis

Evaluations

  • AA-Briefcase1,294#22
  • AA-Omniscience Accuracy18.2%#136
  • AA-Omniscience Non-Hallucination74.2%#4
  • AutomationBench-AA37.2%#24
  • GDPval-AA1,359#24
  • Humanity's Last Exam32%#79
  • GDP.pdf7.4%#27
  • Terminal-Bench 4.01.5%#72
  • AA-LCR80%#44
  • SciCode42.9%#93
  • CritPt4.6%#84
  • Terminal-Bench-Science0%#25
  • FrontierCode32.1%#23
  • MLCR-AA12.2%#20
  • PersonalityBench: Expressiveness21.3#23
  • PersonalityBench: Organization44.2#19
  • PersonalityBench: Transparency13.7#26
  • PersonalityBench: Slop94.9#1
  • PersonalityBench: Informativeness69.5#10
  • PersonalityBench: Warmth13.7#25
  • PersonalityBench: Conviction11.2#26
  • PersonalityBench: Imagination46.7#18
  • GPQA Diamond87.3%#37
  • τ²-Bench Banking34.2%#26
K2 Horizon 375B A23BMedian
Data from Artificial Analysis

Cost per task

Time per task

seconds per task · #1 of 26
  1. 5555K2 Horizon 375B A23B
  2. 5757Beam 1.0
  3. 9696Nemotron 3.5 Super VL
  4. 165165Inkling (Xhigh)
  5. 263263Step 5 Preview
  6. 284284Muse Spark Open (Xhigh)
  7. 341341Muse Wonder (Max)
  8. 364364GPT-6 Luna (Max)
  9. 396396DeepSeek V4.1 Flash 1017 (Max)
  10. 412412GPT-6 Astra (Max)
Data from Artificial Analysis

Tokens per task

output tokens per task · #8 of 27
  1. 34.7K35KInkling (Xhigh)
  2. 38.1K38KGPT-6.1 Sol (Max)
  3. 40.4K40KMuse Spark Open (Xhigh)
  4. 49.2K49KNemotron 3.5 Super VL
  5. 50K50KGPT-6 Luna (Max)
  6. 51.5K52KK2 Horizon 375B A23B
  7. 53.8K54KMuse Wonder Full (Max)
  8. 54.5K55KMuse Wonder (Max)
  9. 55.2K55KKimi K3.1 Pro
  10. 61.6K62KGemini 4 Argon (High)
Data from Artificial Analysis

Speed and price

Speed

tokens per second · #2 of 162
  1. 1,6191,619Celeris-1
  2. 879879K2 Horizon 375B A23B
  3. 714714Mercury 2.5
  4. 446446Nemotron 3.5 Super VL
  5. 414414Gemini 3.5 Flash-Lite
  6. 363363Beam 1.0
  7. 353353Trinity Large Thinking
  8. 295295Nemotron 3.5 Lightning
  9. 290290Nova Micro
  10. 277277NVIDIA Nemotron 3 Nano 30B A3B (Non-reasoning)
Data from Artificial Analysis

Latency

seconds to first token · #85 of 162
  1. 2.12.1GPT-6.1 Sol (Low)
  2. 2.12.1Qwen3.6 35B A3B (Reasoning)
  3. 2.12.1Qwen3.8-Flash-Next
  4. 2.22.2Qwen3.5 397B A17B (Non-reasoning)
  5. 2.22.2Qwen3.5 397B A17B (Reasoning)
  6. 2.22.2K2 Horizon 375B A23B
  7. 2.22.2GPT-5.6 Terra (High)
  8. 2.22.2Gemma 4 12B (Reasoning)
  9. 2.32.3Gemma 4 12B (Non-reasoning)
  10. 2.32.3Qwen3.5 122B A10B (Reasoning)
Data from Artificial Analysis

Price

$ per 1M tokens (3:1) · #1 of 169
  1. $00K2 Horizon 375B A23B
  2. $00Gemma 4 31B (Reasoning)
  3. $00Command A+
  4. $00Ling 3.0 Tiny
  5. $00North Mini Code
  6. $00Tiny Aya Global
  7. $0.040.04Gemma 4 E4B (Reasoning)
  8. $0.040.04Gemma 4 E4B (Non-reasoning)
  9. $0.050.05Granite 4.2 3B
  10. $0.060.06Qwen3.5 4B (Reasoning)
Data from Artificial Analysis

Nearby models

  • Qwen3.8 27B (Xhigh)Alibaba33.7
  • Motif 3Motif Technologies33.6
  • Gemini 3.8 Flash (Low)Google33.5
  • GPT-6 Luna (High)OpenAI32.9
  • GPT-5.3 Codex (Xhigh)OpenAI32.5
  • Beam 1.0Reflection30.3
  • GPT-5.6 Terra (Medium)OpenAI30.1
  • GPT-6 Luna (Medium)OpenAI29.9
Data from Artificial Analysis

Same lab

  • K2 Horizon MoVA 36B A4BInstitute of Foundation Models25.3
  • K2 Horizon 7BInstitute of Foundation Models20.6
  • K2 Horizon 3.7BInstitute of Foundation Models15.6
  • K2 Think V2Institute of Foundation Models11.5
  • K2-V2 (High)Institute of Foundation Models9.9
  • K2-V2 (Medium)Institute of Foundation Models9
  • K2-V2 (Low)Institute of Foundation Models7.3
  • K2 Horizon 0.9BInstitute of Foundation Models3
Data from Artificial Analysis
HorizonVascend IDEOlympusTalosCoursesCommunityBenchmark
DiscordGitHubXLet’s talkLegalPrivacyCookiesData useSecurity
© 2026 Vascend Technologies · VAT IT04911550616
ascend