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

MiniMax-M3

  • MiniMax
  • Release · June 1, 2026
  • Open weights
  • Reasoning

Intelligence

Intelligence Index

#74 of 266
  1. 30.330Beam 1.0
  2. 30.130GPT-5.6 Terra (Medium)
  3. 29.930GPT-6 Luna (Medium)
  4. 29.730Gemini 3.1 Pro Preview
  5. 29.530Claude Haiku 5.5 (Low)
  6. 29.229MiniMax-M3
  7. 28.228Nex-N2-Pro (Based on Qwen3.5-397B-A17B)
  8. 27.628Qwen3.8 27B (Medium)
  9. 27.528GPT-5.6 Terra (Low)
  10. 27.327JT-4.1 Flash 236B A21B (Non-reasoning)
Data from Artificial Analysis

Evaluations

  • AA-Omniscience Accuracy16.7%#153
  • Humanity's Last Exam39%#53
  • Terminal-Bench 4.02%#71
  • AA-LCR83%#18
  • SciCode47.1%#79
  • CritPt3.7%#88
  • AA-AnalystAgent10%#24
  • MMMU-Pro78.6%#35
  • GPQA Diamond92.9%#12
  • IFBench82.9%#1
  • τ²-Bench Telecom88.9%#16
  • τ²-Bench Banking15.3%#57
  • Terminal-Bench Hard42.4%#10
MiniMax-M3Median
Data from Artificial Analysis

Speed and price

Speed

tokens per second · #88 of 162
  1. 106106Claude Sonnet 5.5 (Medium)
  2. 106106Claude Sonnet 5.5 (High)
  3. 105105Claude Sonnet 5.5 (Low)
  4. 105105Muse Glimmer (High)
  5. 105105DeepSeek V4 Pro 0813 (Max)
  6. 103103MiniMax-M3
  7. 103103Qwen3 Omni 30B A3B (Reasoning)
  8. 100100DeepSeek V4 Pro 0813 (Non-reasoning)
  9. 9999GPT-5.3 Codex (Xhigh)
  10. 9898GLM-5.3 (Max)
Data from Artificial Analysis

Latency

seconds to first token · #52 of 162
  1. 1.21.2Step 5 Preview
  2. 1.21.2Nemotron 3 Ultra 550B A55B (Reasoning)
  3. 1.21.2DeepSeek V4.1 Flash 1017 (Max)
  4. 1.31.3Quasar 438B (Max, Based on GLM-5.2)
  5. 1.31.3Trinity Large Thinking
  6. 1.31.3MiniMax-M3
  7. 1.31.3Llama 3.2 Instruct 11B (Vision)
  8. 1.41.4MiniMax-M3.1-Flash
  9. 1.41.4Claude Sonnet 5.5 (Medium)
  10. 1.41.4GPT-5.6 Terra (Low)
Data from Artificial Analysis

Price

$ per 1M tokens (3:1) · #77 of 169
  1. $0.430.43Qwen3 Omni 30B A3B Instruct
  2. $0.450.45Ling 3.1 Flash 1019
  3. $0.450.45Ling 3.1 Flash
  4. $0.480.48Qwen3 Coder Next
  5. $0.530.53DeepSeek V4.1 Flash 1017 (Max)
  6. $0.530.53MiniMax-M3
  7. $0.530.53Solar Pro 4
  8. $0.530.53Inkling Small
  9. $0.530.53Muse Glimmer (High)
  10. $0.540.54MiMo-V2.6-Pro
Data from Artificial Analysis

Nearby models

  • Beam 1.0Reflection30.3
  • GPT-5.6 Terra (Medium)OpenAI30.1
  • GPT-6 Luna (Medium)OpenAI29.9
  • Gemini 3.1 Pro PreviewGoogle29.7
  • Claude Haiku 5.5 (Low)Anthropic29.5
  • Nex-N2-Pro (Based on Qwen3.5-397B-A17B)Nex AGI28.2
  • Qwen3.8 27B (Medium)Alibaba27.6
  • GPT-5.6 Terra (Low)OpenAI27.5
Data from Artificial Analysis

Same lab

  • MiniMax-M3.1-FlashMiniMax42.7
Data from Artificial Analysis
HorizonVascend IDEOlympusTalosCoursesCommunityBenchmark
DiscordGitHubXLet’s talkLegalPrivacyCookiesData useSecurity
© 2026 Vascend Technologies · VAT IT04911550616
ascend