GitHub[225K]Try OpenCode

Compare AI models: DeepSeek V4 Flash vs Step 5 Preview

  1. DeepSeek V4 FlashDeepSeek
  2. Step 5 PreviewStepFunHollow points use fallbacks · hover for details
ReasoningCodingCost efficiencyContext windowMultimodalTool use
Normalized model capability scores
ModelReasoningCodingCost efficiencyContext windowMultimodalTool use
DeepSeek V4 Flash26/10073/10053/10081/1000/1000/100
Step 5 Preview50/100 — Reasoning supported; no comparable benchmark50/100 — No data; neutral placeholder36/10081/10050/10050/100 — Tool calling supported; no comparable benchmark
Overview
Author
Context length
1M
1M
Reasoning
True
True
Input modalities
Text
Text, Image, Video
Output modalities
Text
Text
Providers
Pricing
Input
$0.44/ 1M
$1.05/ 1M
Output
$1.32/ 1M
$2.99/ 1M
Cached input
Unknown
$0.10/ 1M
MomentumLast 2 mo
Unique users
1.1M
124K
Completed sessions
17,144,953
2,664,558
Token share
3.88%
0.54%
Tokens
52T-82%
7.2T+100%
SEP 12NOV 6
SEP 12NOV 6
RetentionWeek 1
Returning users
60.3%88K user-weeks
59.2%2.2K user-weeks

Related comparisons. Other model pairs to check.