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

Step3 VL 10B

  • StepFun
  • Open weights
  • Reasoning

Intelligence

Intelligence Index

#198 of 266
  1. 8.28.2Nova 2.0 Omni (Non-reasoning)
  2. 7.97.9Hermes 4 - Llama-3.1 70B (Reasoning)
  3. 7.87.8Falcon-H1R-7B
  4. 7.87.8Qwen3 Omni 30B A3B (Reasoning)
  5. 7.87.8Gemma 4 E2B (Reasoning)
  6. 7.77.7Step3 VL 10B
  7. 7.77.7Llama 3.3 Instruct 70B
  8. 7.57.5Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)
  9. 7.57.5ERNIE 4.5 300B A47B
  10. 7.57.5NVIDIA Nemotron Nano 12B v2 VL (Reasoning)
Data from Artificial Analysis

Evaluations

  • AA-Omniscience Accuracy13.3%#195
  • Humanity's Last Exam10.8%#146
  • AA-LCR0%#229
  • CritPt0%#143
  • MMMU-Pro64%#79
  • GPQA Diamond69%#122
  • IFBench50.2%#63
  • τ²-Bench Telecom16.1%#121
  • Terminal-Bench Hard5.3%#79
Step3 VL 10BMedian
Data from Artificial Analysis

Nearby models

  • Nova 2.0 Omni (Non-reasoning)Amazon8.2
  • Hermes 4 - Llama-3.1 70B (Reasoning)Nous Research7.9
  • Falcon-H1R-7BTII UAE7.8
  • Qwen3 Omni 30B A3B (Reasoning)Alibaba7.8
  • Gemma 4 E2B (Reasoning)Google7.8
  • Llama 3.3 Instruct 70BMeta7.7
  • Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)NVIDIA7.5
  • ERNIE 4.5 300B A47BBaidu7.5
Data from Artificial Analysis

Same lab

  • Step 5 PreviewStepFun43.7
Data from Artificial Analysis
HorizonVascend IDEOlympusTalosCoursesCommunityBenchmark
DiscordGitHubXLet’s talkLegalPrivacyCookiesData useSecurity
© 2026 Vascend Technologies · VAT IT04911550616
ascend