Key Specifications
| Vendor | google |
|---|
| Version | 1.0-ultra |
|---|
| Release Date | 2023-12-06 |
|---|
| Context Window | 32768 tokens |
|---|
| Input Modalities | text, image |
|---|
| Output Modalities | text |
|---|
| License | Proprietary |
|---|
| Documentation | https://ai.google.dev/gemini-api/docs |
|---|
Benchmark Performance
| Benchmark | Score | Unit | Evaluated At | Notes | Source |
|---|
| MMLU | 80.3 | % | 2023-12-06 | 5-shot | view |
| HUMANEVAL | 81.7 | pass@1 | 2023-12-06 | — | view |
| GSM8K | 85.9 | % | 2023-12-06 | 0-shot CoT | view |
| MATH | 57.1 | % | 2023-12-06 | 0-shot CoT | view |
| BBH | 83.8 | % | 2023-12-06 | 3-shot CoT | view |
| GPQA | 46.9 | % | 2023-12-06 | 0-shot | view |
| IFEVAL | 81.6 | % | 2023-12-06 | prompt_strict | view |
| ARC | 95.7 | % | 2023-12-06 | challenge | view |
| MUSR | 64.3 | % | 2023-12-06 | 0-shot | view |
| WINOGRANDE | 80.4 | % | 2023-12-06 | 0-shot | view |
Pricing
| Tier | Price | Currency |
|---|
| Input | $7 / Mtok | USD |
| Output | $21 / Mtok | USD |
| Cache Read | $0 / Mtok | USD |
| Cache Write | $0 / Mtok | USD |
Source:
https://ai.google.dev/pricing
· as of 2023-12-06
Compliance
- Data Residency: US
- SOC2: ✓
- HIPAA: ✗
- GDPR: ✓
- ISO 27001: ✓
Gemini 1.0 Ultra
Modeloverzicht
Google Gemini 1.0 Ultra 旗舰模型, 32K 上下文, 在多项基准上接近 GPT-4, 多模态推理能力突出。
Kernspecificaties
| Leverancier | Versie | Releasedatum | Contextvenster | Invoermodaliteiten | Uitvoermodaliteiten | Licentie |
|---|
| Google | 1.0-ultra | 2023-12-06 | 32K | text, image | text | Proprietary |
Benchmarkprestaties
| Benchmark | Score | Eenheid | Notities |
|---|
| MMLU (Massive Multitask Language Understanding) | 80.3 | % | 5-shot |
| HumanEval | 81.7 | pass@1 | — |
| GSM8K (Grade School Math 8K) | 85.9 | % | 0-shot CoT |
| MATH | 57.1 | % | 0-shot CoT |
| BBH (BIG-Bench Hard) | 83.8 | % | 3-shot CoT |
| GPQA | 46.9 | % | 0-shot |
| IFEval | 81.6 | % | prompt_strict |
| ARC | 95.7 | % | challenge |
| MUSR | 64.3 | % | 0-shot |
| WinoGrande | 80.4 | % | 0-shot |
Prijzen
| Invoer | Uitvoer | Cache-lezen | Cache-schrijven |
|---|
| — | — | — | — |
per miljoen tokens
Sterktes
- MMLU score 80.3, strong knowledge reasoning.
- HumanEval 81.7, excellent code generation.
- GSM8K 85.9, robust math reasoning.
- 支持文本、图像、音频多模态输入。
Zwaktes
Gebruiksscenario’s
- 代码生成与调试
- 视觉与图像理解
- Agent 工作流与工具调用
Referenties