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

Mistral Small 4 (Reasoning)

  • Mistral
  • Release · March 16, 2026
  • Open weights
  • Reasoning

Intelligence

Intelligence Index

#153 of 266
  1. 11.712Nemotron Cascade 2 30B A3B
  2. 11.612gpt-oss-120b (High)
  3. 11.512K2 Think V2
  4. 11.512LongCat Flash Lite
  5. 11.411HyperCLOVA X SEED Think (32B)
  6. 11.311Mistral Small 4 (Reasoning)
  7. 11.211Qwen3.5 9B (Reasoning)
  8. 11.211Qwen3 Next 80B A3B (Reasoning)
  9. 11.111Granite 4.2 8B
  10. 11.111Ling 3.0 Tiny
Data from Artificial Analysis

Evaluations

  • AA-Omniscience Accuracy21.7%#109
  • Humanity's Last Exam9.9%#156
  • Terminal-Bench 4.00%#96
  • AA-LCR49.7%#147
  • SciCode38.8%#111
  • CritPt0.3%#117
  • AA-AnalystAgent1.3%#26
  • MMMU-Pro56.8%#93
  • GPQA Diamond76.9%#80
  • IFBench48.2%#66
  • τ²-Bench Telecom41.2%#74
  • τ²-Bench Banking5%#89
  • Terminal-Bench Hard17.4%#42
Mistral Small 4 (Reasoning)Median
Data from Artificial Analysis

Speed and price

Speed

tokens per second · #31 of 162
  1. 186186Claude Haiku 5.5 (Xhigh)
  2. 183183Nova 2.0 Lite (Non-reasoning)
  3. 181181Claude Haiku 5.5 (High)
  4. 180180Nova 2.0 Lite (Medium)
  5. 179179Claude Haiku 5.5 (Medium)
  6. 178178Mistral Small 4 (Reasoning)
  7. 177177DeepSeek V4 Flash Vision (Max)
  8. 173173Mistral Small 4 (Non-reasoning)
  9. 173173Claude Haiku 5.5 (Low)
  10. 173173gpt-oss-20b (High)
Data from Artificial Analysis

Latency

seconds to first token · #25 of 162
  1. 0.80.8Nova Micro
  2. 0.80.8Granite 4.2 30B
  3. 0.80.8Nemotron 3 Nano Omni 30B A3B Reasoning
  4. 0.90.9Phi-4 Mini Instruct
  5. 0.90.9Ministral 3 8B
  6. 0.90.9Mistral Small 4 (Reasoning)
  7. 0.90.9GPT-6 Luna (Non-reasoning)
  8. 0.90.9GPT-5.5 Instant (June 2026)
  9. 0.90.9Qwen3.5 9B (Non-reasoning)
  10. 0.90.9Qwen3.5 4B (Non-reasoning)
Data from Artificial Analysis

Price

$ per 1M tokens (3:1) · #57 of 169
  1. $0.240.24GLM-5.3-Flash
  2. $0.240.24Hy3
  3. $0.240.24gpt-oss-120b (Low)
  4. $0.260.26MiniMax-M3.1-Flash
  5. $0.260.26gpt-oss-120b (High)
  6. $0.260.26Mistral Small 4 (Reasoning)
  7. $0.260.26Mistral Small 4 (Non-reasoning)
  8. $0.280.28Qwen3.5 Omni Flash
  9. $0.280.28Granite 4.2 30B
  10. $0.30.3Llama 4 Scout
Data from Artificial Analysis

Nearby models

  • Nemotron Cascade 2 30B A3BNVIDIA11.7
  • gpt-oss-120b (High)OpenAI11.6
  • K2 Think V2Institute of Foundation Models11.5
  • LongCat Flash LiteLongCat11.5
  • HyperCLOVA X SEED Think (32B)Naver11.4
  • Qwen3.5 9B (Reasoning)Alibaba11.2
  • Qwen3 Next 80B A3B (Reasoning)Alibaba11.2
  • Granite 4.2 8BIBM11.1
Data from Artificial Analysis

Same lab

  • Mistral Large 4 PreviewMistral38.4
  • Mistral Medium 3.5Mistral14.2
  • Magistral Medium 1.2Mistral11.8
  • Mistral Small 4 (Non-reasoning)Mistral9
  • Magistral Small 1.2Mistral8.6
  • Ministral 3 14BMistral6
  • Ministral 3 8BMistral5.5
  • Ministral 3 3BMistral4.8
Data from Artificial Analysis
HorizonVascend IDEOlympusTalosCoursesCommunityBenchmark
DiscordGitHubXLet’s talkLegalPrivacyCookiesData useSecurity
© 2026 Vascend Technologies · VAT IT04911550616
ascend