Llama 3.1 Instruct: pricing, context window and benchmarks
Llama 3.1 Instruct is a multimodal model from Meta AI. It offers a 200,000-token context window, costs $4.05 per million input tokens and $16.24 per million output tokens, and averages 75.7 across the index's benchmark suite.
Pricing
Llama 3.1 Instruct costs $4.05 per million input tokens and $16.24 per million output tokens. A blended 3:1 input-to-output mix works out to about $7.10 per million tokens. Cached input is listed at $0.40 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $4.05 | per 1M tokens |
| Output | $16.24 | per 1M tokens |
| Cached input | $0.40 | per 1M tokens |
| Blended 3:1 | $7.10 | estimate per 1M tokens |
| 1M in + 1M out | $20.29 | 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 | 79.5 | 196 | 67th | |
| GPQA Diamond | 71.8 | 223 | 63th | |
| SWE-bench Verified | 64.2 | 229 | 62th | |
| HumanEval | 81.2 | 209 | 65th | |
| MATH-500 | 77.0 | 278 | 54th | |
| AIME 2025 | 72.0 | 220 | 63th | |
| LiveCodeBench | 67.0 | 252 | 58th | |
| IFEval | 85.5 | 174 | 71th | |
| TruthfulQA | 79.0 | 217 | 64th | |
| HellaSwag | 84.9 | 252 | 58th | |
| MMMU | 70.9 | 274 | 54th | |
| Arena Elo | 1395.5 | 226 | 62th |
What Llama 3.1 Instruct is good for
Llama 3.1 Instruct 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 $4.05 per million input tokens (hosted list price) and a 200,000-token context window, it is a practical option for teams that want more control than a closed API allows.
Strengths
- Large 200,000-token context window for long documents and agent traces.
- 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 |
|---|---|---|---|---|
| Hunyuan-Turbo | Tencent | 128000 | $10.55 | 75.8 |
| Llama 3 Lambda Instruct | Lambda | 32768 | $3.92 | 75.8 |
| Llama 3 Anyscale Instruct | Anyscale | 200000 | $3.91 | 75.7 |
| Llama 3 Replicate | Replicate | 200000 | $3.75 | 75.7 |
| Llama 3.1 Instruct | Meta AI | 200000 | $4.05 | 75.7 |
| Sonar Turbo | Perplexity | 65536 | $10.71 | 75.7 |
FAQ
How much does Llama 3.1 Instruct cost?
Llama 3.1 Instruct is listed at $4.05 per million input tokens and $16.24 per million output tokens. Cached input is $0.40 per million tokens. A blended 3:1 input-to-output mix is about $7.10 per million tokens.
What is the context window of Llama 3.1 Instruct?
Llama 3.1 Instruct has a context window of 200,000 tokens and can emit up to 8,192 output tokens per request in this index.
Is Llama 3.1 Instruct open weights?
Yes. Llama 3.1 Instruct is listed as an open-weight model under a open weights licence.
How does Llama 3.1 Instruct score on benchmarks?
Llama 3.1 Instruct averages 75.7 across the scored benchmarks in this index, ranking #232 of 600 models. Its strongest relative result is Arena Elo at 1395.5.
What is Llama 3.1 Instruct best for?
Llama 3.1 Instruct 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 $4.05 per million input tokens (hosted list price) and a 200,000-token context window, it is a practical option for teams that want more control than a closed API allows.