Llama 3.1 Turbo Base: pricing, context window and benchmarks
Llama 3.1 Turbo Base is a open-weight model from Together AI. It offers a 16,384-token context window, costs $3.30 per million input tokens and $13.59 per million output tokens, and averages 73.0 across the index's benchmark suite.
Pricing
Llama 3.1 Turbo Base costs $3.30 per million input tokens and $13.59 per million output tokens. A blended 3:1 input-to-output mix works out to about $5.87 per million tokens. Cached input is listed at $0.47 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $3.30 | per 1M tokens |
| Output | $13.59 | per 1M tokens |
| Cached input | $0.47 | per 1M tokens |
| Blended 3:1 | $5.87 | estimate per 1M tokens |
| 1M in + 1M out | $16.89 | 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 | 75.8 | 310 | 48th | |
| GPQA Diamond | 65.6 | 350 | 42th | |
| SWE-bench Verified | 62.2 | 275 | 54th | |
| HumanEval | 79.4 | 252 | 58th | |
| MATH-500 | 76.8 | 284 | 53th | |
| AIME 2025 | 68.3 | 290 | 52th | |
| LiveCodeBench | 66.5 | 264 | 56th | |
| IFEval | 79.6 | 342 | 43th | |
| TruthfulQA | 77.2 | 275 | 54th | |
| HellaSwag | 84.5 | 267 | 56th | |
| MMMU | 67.0 | 358 | 40th | |
| Arena Elo | 1381.0 | 271 | 55th |
What Llama 3.1 Turbo Base is good for
Llama 3.1 Turbo Base 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 $3.30 per million input tokens (hosted list price) and a 16,384-token context window, it is a practical option for teams that want more control than a closed API allows.
Strengths
- Input price of $3.30 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
- Context window of 16,384 tokens is small for long-document work.
- Average benchmark score is below the index median; test it on your own task.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| Codestral Base | Mistral AI | 8192 | $3.31 | 73.1 |
| Mixtral Groq | Groq | 16384 | $3.37 | 73.1 |
| Claude 3.7 Sonnet Mini | Anthropic | 32768 | $9.74 | 73.0 |
| Llama 3.1 Turbo Base | Together AI | 16384 | $3.30 | 73.0 |
| Nemotron 4 Base | NVIDIA | 8192 | $3.29 | 73.0 |
| Mixtral Instruct | Mistral AI | 65536 | $3.35 | 72.9 |
FAQ
How much does Llama 3.1 Turbo Base cost?
Llama 3.1 Turbo Base is listed at $3.30 per million input tokens and $13.59 per million output tokens. Cached input is $0.47 per million tokens. A blended 3:1 input-to-output mix is about $5.87 per million tokens.
What is the context window of Llama 3.1 Turbo Base?
Llama 3.1 Turbo Base has a context window of 16,384 tokens and can emit up to 16,384 output tokens per request in this index.
Is Llama 3.1 Turbo Base open weights?
Yes. Llama 3.1 Turbo Base is listed as an open-weight model under a open weights licence.
How does Llama 3.1 Turbo Base score on benchmarks?
Llama 3.1 Turbo Base averages 73.0 across the scored benchmarks in this index, ranking #302 of 600 models. Its strongest relative result is Arena Elo at 1381.0.
What is Llama 3.1 Turbo Base best for?
Llama 3.1 Turbo Base 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 $3.30 per million input tokens (hosted list price) and a 16,384-token context window, it is a practical option for teams that want more control than a closed API allows.