modelgrep

OpenAI: o3 vs StepFun: Step 3.5 Flash

StepFun: Step 3.5 Flash wins on more metrics (6 of 9), but the right pick depends on what you optimize for — see the breakdown below.

MetricOpenAI: o3StepFun: Step 3.5 Flash
Intelligence Index38.437.8
Coding Index38.431.6
GPQA Diamond83%83%
Design Arena Elo1200
Speed (tokens/sec)4045
Latency5.4s449ms
Input price /M$2.00$0.090
Output price /M$8.00$0.300
Context window200K262K
CapabilitiesReasoningToolsJSONVisionReasoningToolsJSON