Side-by-side technical and economic comparison between Meta / Together's Llama 4 Scout (17B Active / 109B MoE) — Together AI and Google's Gemini 3.8 Pro.
• Llama 4 Scout (17B Active / 109B MoE) — Together AI: Classified as a Legacy Reference model (Deprecated on Together AI (Reference)).
| Metric | Llama 4 Scout (17B Active / 109B MoE) — Together AI | Gemini 3.8 Pro | Advantage |
|---|---|---|---|
| Provider Organization | Meta / Together | — | |
| Standard Input / 1M | $0.18 | $4.00 | Llama 4 Scout (17B Active / 109B MoE) — Together AI (96% lower) |
| Cached Input / 1M | None | $0.40 | Llama 4 Scout (17B Active / 109B MoE) — Together AI lower |
| Output / 1M | $0.59 | $18.00 | Llama 4 Scout (17B Active / 109B MoE) — Together AI lower |
| Context Window | 328k | 1.05M | Gemini 3.8 Pro (1.05M) |
| Max Generation Tokens | 16.4k | 64k | Gemini 3.8 Pro |
| Tokenizer Family | Meta Llama 3/4 Tiktoken (128k vocabulary) | Google Gemini SentencePiece (256k vocabulary) | — |
Opt for Llama 4 Scout (17B Active / 109B MoE) — Together AI when your engineering requirements prioritize Meta / Together's ecosystem, specific tokenizer efficiencies (Calibrated Llama 3 Tokenizer (±3%)), or when your expected prompt-to-completion ratios favor its $0.18/M input rate.
Opt for Gemini 3.8 Pro when looking for Google's tooling integration, specific context window depth (1.05M tokens), or when output generation volume favors its $18/M rate.