Gemma 4 26B A4B vs Llama 3.3 70B Instruct
Gemma 4 26B A4B is cheaper on output tokens, while Gemma 4 26B A4B offers a larger context window. Choose Gemma 4 26B A4B or Llama 3.3 70B Instruct based on the trade-off between cost, context, and the benchmarks that matter for your use case.
| Spec | Gemma 4 26B A4B | Llama 3.3 70B Instruct |
|---|---|---|
| Provider | Meta | |
| Input / 1M tokens | $0.09 | $0.10 |
| Output / 1M tokens | $0.30 | $0.32 |
| Context window | 262K | 131K |
| Parameters | — | — |
| Open weights | Yes | — |
| Released | Apr 2026 | — |
| LMArena (Chatbot Arena) Elo score | 1434 | 1274 |
| LMArena WebDev Elo score | 1360 | — |
| Epoch Capabilities Index score | 142 | — |
| LMArena Vision Elo score | 1260 | — |
| GPQA Diamond score | — | 47.4 |
FAQ
Is Gemma 4 26B A4B or Llama 3.3 70B Instruct cheaper?
Gemma 4 26B A4B is cheaper on output tokens ($0.30 vs $0.32 per 1M).
Which has the larger context window, Gemma 4 26B A4B or Llama 3.3 70B Instruct?
Gemma 4 26B A4B has the larger context window (262K tokens).
Pricing is indicative — confirm with the provider before production use. Highlighted values indicate the better figure for that row.