Massive 405-billion parameter open-weights flagship rivaling the largest proprietary models
meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo| Scenario | Input Tokens | Output Tokens | Uncached Cost | With Prompt Caching |
|---|---|---|---|---|
| Short Chat Query | 1,000 | 500 | $0.00525 | N/A |
| Document Summarization | 10,000 | 2,000 | $0.0420 | N/A |
| Codebase & Context Analysis | 100,000 | 20,000 | $0.4200 | N/A |
| Batch Corpus Processing | 1,000,000 | 100,000 | $3.85 | N/A |
Ideal for production workloads demanding flagship capabilities, deep context depth (128k tokens), and reliability from Meta. Excellent when predictable tokenomics and prompt caching support are paramount.
If your use-case requires sub-second streaming latency or ultra-high frequency classification at micro-cent pricing, consider lighter budget options such as Gemini Flash-Lite or Claude Haiku. For deep formal logic, consider dedicated reasoning models like o3.
For Llama 3.1 405B — Together AI, 1 million input tokens costs $3.50, while 1 million output tokens costs $3.50. If using prompt caching, repetitive input prefixes are discounted to $0.875 per million.
Llama 3.1 405B — Together AI features a maximum context window of 131,072 tokens (~98,304 words), with a maximum output limit of 4,096 tokens per completion.
Llama 3.1 405B — Together AI utilizes the Meta Llama 3 Tiktoken BPE (~128k vocabulary). Token counting on TokenMath runs client-side to ensure maximum privacy.
No, Llama 3.1 405B — Together AI does not currently advertise prompt caching discounts on its standard API tier.