Compare models
Put up to 4 models beside each other — token prices, context windows, capabilities and provider, from the same catalogue the model pages read.
| Attribute | Universal-3.5 Prouniversal-3-5-pro | Nemotron 3.5 Lightningnemotron-3.5-lightning | Gemini 4 Argongemini-4-argon |
|---|---|---|---|
| Pricing | |||
| Input | — Not priced per input token | — Not priced per input token | $2.00 / 1M |
| Output | — Not priced per output token | — Not priced per output token | $10.00 / 1M |
| Cache Write (5m) | Not applicable | $0.00 / 1M | $2.00 / 1M |
| Cache Write (1h) | Not applicable | $0.00 / 1M | $2.00 / 1M |
| Cache Read | Not applicable | $0.00 / 1M | $2.00 / 1M |
| Web Search | $0 / 1M | $0 / 1M | $0 / 1M |
| Context | |||
| Max context | N/A | 1M | 1M |
| Max output | N/A | N/A | N/A |
| Capabilities | |||
| Vision | No | Yes | Yes |
| Function Calling | No | Yes | Yes |
| JSON Mode | No | Yes | Yes |
| Streaming | No | Yes | Yes |
| Catalogue | |||
| Provider | AssemblyAI | NVIDIA | |
| Category | voice | chat | chat |
| Charge type | Pay As You Go | Pay As You Go | Pay As You Go |
| Released | 2026-09-22 | — | — |
| Description | |||
| Summary | AssemblyAI Universal-3.5 Pro speech-to-text. Billed per second of audio. | NVIDIA Nemotron 3.5 Lightning is an open Mixture-of-Experts (MoE) model with 30B total parameters and 3B active per token, optimized for high-throughput agentic workloads and efficient inference. Its lightweight active compute and open design make it well suited for specialized agents, domain-specific customization, and scalable production deployments where speed, cost efficiency, and adaptability are key. | Gemini 4 Argon is Google's flagship reasoning model, built for complex software engineering, deep mathematical and scientific problem solving, and autonomous multi-step agents. Google trained it to keep context across long, multi-step tasks and to turn long documents mixing text, tables, charts and images into a single grounded output. It leads the Vals Index across finance, legal, coding and tax work, and supports long decode of up to 1M output tokens in a single response. |