Models · Compare
Llama 4 Scout vs Gemini 2.5 Pro
Rows highlighted in warm gray are where the models differ. Numbers carry their as-of date and primary source.
Specs
| Field | A: Llama 4 Scout | B: Gemini 2.5 Pro |
|---|---|---|
| Released | 2025-04-05 | 2025-03-25 |
| Developer | Meta | Google DeepMind |
| Openness | Source-available | Proprietary |
| License | Llama 4 Community License | Proprietary |
| OSI-approved | no | no |
| Data released | no | no |
| Training code | no | no |
| Architecture | moe | unknown |
| Total params | 109B | — |
| Active params | 17B | — |
| Experts | 16 (1 active) | — |
| Context window | 10.5M | 1.0M |
| Attention | gqa | unknown |
| Position enc. | rope-llama3 | unknown |
| Pretraining tokens | — | — |
| Post-training | sft, dpo, rejection-sampling | rlhf |
| Training hardware | H100 | — |
| $/M input | $0.17 | $1.25 |
| $/M output | $0.66 | $10.00 |
| Output tok/sec | 108.1 | 127.3 |
Benchmarks
Missing scores render as not reported; never inferred. Bold highlights the leader per benchmark.
General reasoning
| MMLU-Pro | — | 86.2 2026-05-21 |
Code
| LiveCodeBench | 29.9 2026-05-21 | 80.1 2026-05-21 |
Math
| MATH | 84.4 2026-05-21 | 96.7 2026-05-21 |
| AIME 2024 | 28.3 2026-05-21 | — |
| AIME 2025 | 14.0 2026-05-21 | 87.7 2026-05-21 |
Held-out / arena
Context · A
Llama's first MoE family. Scout's headline was a 10M-token context window and natively multimodal vision-language input. The MoE pivot followed DeepSeek V3 in establishing MoE as mainstream for open-weights frontier work.
Context · B
The first Gemini release to clearly lead on LMArena Elo and on hard reasoning benchmarks. Native 1M-token context with reported 2M expansion in pipeline.