Models · Compare
Phi-4 vs Gemini 2.0 Flash
Rows highlighted in warm gray are where the models differ. Numbers carry their as-of date and primary source.
Specs
| Field | A: Phi-4 | B: Gemini 2.0 Flash |
|---|---|---|
| Released | 2024-12-12 | 2024-12-11 |
| Developer | Microsoft | Google DeepMind |
| Openness | Open | Proprietary |
| License | MIT | Proprietary |
| OSI-approved | yes | no |
| Data released | no | no |
| Training code | no | no |
| Architecture | dense | unknown |
| Total params | 14B | — |
| Active params | — | — |
| Experts | — | — |
| Context window | 16K | 1.0M |
| Attention | gqa | unknown |
| Position enc. | rope | unknown |
| Pretraining tokens | 9.8T | — |
| Post-training | sft, dpo | rlhf |
| Training hardware | — | — |
| $/M input | $0.13 | $0.15 |
| $/M output | $0.50 | $0.60 |
| Output tok/sec | 30.9 | 0 |
Benchmarks
Missing scores render as not reported; never inferred. Bold highlights the leader per benchmark.
General reasoning
| MMLU | 84.8 2024-12-12 | — |
| MMLU-Pro | 71.4 2026-05-21 | 77.9 2026-05-21 |
| GPQA-Diamond | 56.1 2024-12-12 | 62.3 2026-05-21 |
Code
| HumanEval | 82.6 2024-12-12 | — |
| LiveCodeBench | 23.1 2026-05-21 | 33.4 2026-05-21 |
Math
| MATH | 80.4 2024-12-12 | 93.0 2026-05-21 |
| AIME 2024 | 14.3 2026-05-21 | 33.0 2026-05-21 |
| AIME 2025 | 18.0 2026-05-21 | 21.7 2026-05-21 |
Context · A
14B parameters reaching benchmark scores within range of much larger models, via aggressive synthetic-data pretraining and data quality curation. The Phi line continues the "data quality over scale" thesis.
Context · B
Google's workhorse Gemini 2.0 checkpoint, released initially as an experimental developer preview. Google positioned it as outperforming Gemini 1.5 Pro on key benchmarks at roughly twice the speed, with native multimodal output (image generation, text-to-speech) and native tool use including Google Search and code execution. The Multimodal Live API added real-time audio and video streaming.