Gemma 3: pricing, context window and benchmarks
Gemma 3 is a multimodal model from Google DeepMind. It offers a 65,536-token context window, costs $3.68 per million input tokens and $14.67 per million output tokens, and averages 73.6 across the index's benchmark suite.
Pricing
Gemma 3 costs $3.68 per million input tokens and $14.67 per million output tokens. A blended 3:1 input-to-output mix works out to about $6.43 per million tokens. Cached input is listed at $0.39 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $3.68 | per 1M tokens |
| Output | $14.67 | per 1M tokens |
| Cached input | $0.39 | per 1M tokens |
| Blended 3:1 | $6.43 | estimate per 1M tokens |
| 1M in + 1M out | $18.35 | estimate |
Benchmarks
Scores below are modelled estimates on a 0–100 scale. Use them to compare relative position, not as official vendor results.
| Benchmark | Score | Rank | Percentile | Relative |
|---|---|---|---|---|
| MMLU-Pro | 75.1 | 330 | 45th | |
| GPQA Diamond | 71.5 | 230 | 62th | |
| SWE-bench Verified | 60.0 | 328 | 45th | |
| HumanEval | 76.6 | 315 | 48th | |
| MATH-500 | 75.7 | 309 | 49th | |
| AIME 2025 | 69.2 | 274 | 54th | |
| LiveCodeBench | 65.4 | 285 | 53th | |
| IFEval | 82.0 | 279 | 54th | |
| TruthfulQA | 78.6 | 232 | 61th | |
| HellaSwag | 83.7 | 290 | 52th | |
| MMMU | 72.1 | 239 | 60th | |
| Arena Elo | 1371.5 | 312 | 48th |
What Gemma 3 is good for
Gemma 3 is an open-weight model, so it can be self-hosted or run through a third-party inference provider. That matters for data residency, fine-tuning and long-run cost control. At $3.68 per million input tokens (hosted list price) and a 65,536-token context window, it is a practical option for teams that want more control than a closed API allows.
Strengths
- Input price of $3.68 per million tokens is at or below the index median.
- Open weights make self-hosting, fine-tuning and data-residency review possible.
- Accepts more than text, including image input where the provider supports it.
Trade-offs
- No standout weakness in the indexed specifications, but real-world behaviour still needs testing.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| Mixtral | Mistral AI | 65536 | $3.48 | 73.7 |
| Gemma 3 | Google DeepMind | 65536 | $3.68 | 73.6 |
| GPT-5.2 Nano | OpenAI | 128000 | $9.73 | 73.5 |
| Llama 3.3 Groq Base | Groq | 16384 | $3.24 | 73.5 |
| Yi-Large | 01.AI | 32768 | $3.40 | 73.5 |
| Yi-Large Chat | 01.AI | 65536 | $3.49 | 73.5 |
FAQ
How much does Gemma 3 cost?
Gemma 3 is listed at $3.68 per million input tokens and $14.67 per million output tokens. Cached input is $0.39 per million tokens. A blended 3:1 input-to-output mix is about $6.43 per million tokens.
What is the context window of Gemma 3?
Gemma 3 has a context window of 65,536 tokens and can emit up to 65,536 output tokens per request in this index.
Is Gemma 3 open weights?
Yes. Gemma 3 is listed as an open-weight model under a open weights licence.
How does Gemma 3 score on benchmarks?
Gemma 3 averages 73.6 across the scored benchmarks in this index, ranking #284 of 600 models. Its strongest relative result is Arena Elo at 1371.5.
What is Gemma 3 best for?
Gemma 3 is an open-weight model, so it can be self-hosted or run through a third-party inference provider. That matters for data residency, fine-tuning and long-run cost control. At $3.68 per million input tokens (hosted list price) and a 65,536-token context window, it is a practical option for teams that want more control than a closed API allows.