Last refreshed 2026-07-09. Next refresh: weekly.
Why use Llama 3.1 405B Instruct on Together AI?
Together AI offers Llama 3.1 405B Instruct with pay-as-you-go pricing at $5.00/1M input tokens. Together AI is a platform for running open-source and proprietary LLMs with fast serverless and dedicated endpoints at competitive inference pricing.
Compare Llama 3.1 405B Instruct across 11 providers to find the best fit for your use caseSetup recipe
Python + curlpip install togetherexport TOGETHER_API_KEY=...from together import Together
client = Together() # reads TOGETHER_API_KEY from env
response = client.chat.completions.create(
model="llama3.1-405b-instruct",llama3.1-405b-instructRequest example
from together import Together
client = Together() # reads TOGETHER_API_KEY from env
response = client.chat.completions.create(
model="llama3.1-405b-instruct",
messages=[{"role": "user", "content": "Hello"}]
)
print(response.choices[0].message.content)Gotchas
- Together uses "organization/model-name" format, e.g. "meta-llama/Llama-4-Scout-17B-16E-Instruct" or "Qwen/QwQ-32B". See the Together model catalog for the exact ID.
- The examples expect TOGETHER_API_KEY; rename it only if your application config maps the new variable.
Compare Llama 3.1 405B Instruct Across Providers
| Provider | Input (per 1M) | Output (per 1M) |
|---|---|---|
| OctoAI API (Deprecated) | — | — |
| Together AI | $5.00 | $15.00 |
| Fireworks AI | $3.00 | $3.00 |
| IBM watsonx | $3.00 | $9.00 |
| Scale AI GenAI Platform | — | — |
Pricing
| Type | Price (per 1M) |
|---|---|
| Input tokens | $5.00 |
| Output tokens | $15.00 |
Capabilities
About Llama 3.1 405B Instruct
Llama 3.1 405B Instruct is Meta's advanced large language model released on July 23, 2024, featuring 405 billion parameters. It utilizes an optimized transformer architecture with supervised fine-tuning and reinforcement learning for enhanced instruction-following capabilities. The model supports multiple languages, was trained on 15 trillion tokens, and fine-tuned with 25 million synthetic examples. It excels in multilingual dialogue and text generation, making it ideal for assistant-like applications. Llama 3.1 incorporates robust safety measures and ethical considerations, outperforming many existing models on various industry benchmarks.