# Llama 3.3 70B Instruct

> Meta's Llama 3.3 70B instruction-tuned model for high-quality chat, reasoning, and coding at a fraction of frontier cost.

**ID**: `meta/llama-3.3-70b-instruct`  
**Creator**: meta  
**Category**: text  
**Context**: 32K tokens  
**Pricing**: $0.10 / $0.32 per 1M · billed by your provider (BYOK)  
**Released**: 2024-12-06  
**Web page**: https://anyrouter.dev/model/meta/llama-3.3-70b-instruct

**Input modalities**: text  
**Output modalities**: text  

**Tokenizer**: Llama

**Capabilities**: chat, reasoning, function-calling, streaming, coding

**Aliases**: `meta-llama/llama-3.3-70b-instruct`, `meta/llama-3.3-70b-instruct-fp8-fast`

## Usage

**Endpoint**: `POST https://anyrouter.dev/api/v1/chat/completions`

```bash
curl https://anyrouter.dev/api/v1/chat/completions \
  -H "Authorization: Bearer $ANYROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "meta/llama-3.3-70b-instruct",
  "messages": [
    {
      "role": "user",
      "content": "Say hi in 3 words."
    }
  ]
}'
```

## Providers

| Provider | Input | Output |
| --- | --- | --- |
| fpt-byok | $0.21/M (list) | $0.45/M (list) |
| huggingface-byok | $0.21/M (list) | $0.45/M (list) |
| openrouter-byok | $0.10/M (list) | $0.32/M (list) |
| aihubmix-byok | $0.60/M (list) | $1.20/M (list) |
| ovhcloud-byok | $0.74/M (list) | $0.74/M (list) |
| novita-byok | $0.14/M (list) | $0.40/M (list) |
| nousresearch-byok | $0.10/M (list) | $0.32/M (list) |

BYOK routes are billed by your own provider key — AnyRouter charges $0. Rates marked (list) are the provider's published list price.

## Benchmarks

- **GPQA Diamond**: 0.5
- **SciCode**: 0.3
- **MATH Level 5**: 0.4
- **OTIS Mock AIME**: 0.1

Benchmark data from [Epoch AI](https://epoch.ai/benchmarks) (CC-BY 4.0).

## Model card

- Total parameters: 70B [1]
- Native context: 131072 tokens [1]
- Training cutoff: 2023-12 [1]
- License: Llama 3.3 Community License Agreement, open weights (https://huggingface.co/meta-llama/Llama-3.3-70B-Instruct) [1]

### Strengths [1]

- Tuned for multilingual dialogue
- Tool use on par with Llama 3.1 70B (BFCL v2 77.3)

### Known limits [1]

- Eight supported languages; others need fine-tuning
- Can give inaccurate, biased or objectionable answers
- Static model trained on an offline dataset

### Use cases [1]

- Assistant-style chat
- Synthetic data generation and distillation
- Multilingual commercial and research use

### Links

- model card: https://github.com/meta-llama/llama-models/blob/main/models/llama3_3/MODEL_CARD.md
- Previous version: https://anyrouter.dev/model/meta/llama-3.1-70b-instruct

### Sources

1. [Llama 3.3 Model Card (meta-llama/llama-models)](https://raw.githubusercontent.com/meta-llama/llama-models/main/models/llama3_3/MODEL_CARD.md) (accessed 2026-10-14)