Compare

NVIDIA logoNemotron 3.5 Content SafetyNVIDIA
Best input$0.20 /MTok
Best output$0.20 /MTok
Batch in / out
Cache read
Measured TTFT
Measured TPS
Context131K
Max output118K
Modalities inimage, text
Tool calling
Structured output
Reasoning
Vision
Audio input
Prompt caching
Providers1
Cheapest providerDeepInfra
LicenseOpen weights
ReleasedJun 4, 2026
Cost at your workload
● = best of the compared models · TTFT/TPS: fastest measured endpoint per model (P50, trailing 21d, chat10k) · ° = via OpenRouter routing · ✕ = probed, no measurable stream · prices are standard tier