Open weights · Meta Llama 3 Community License
Llama 3 70B
Meta's 70B-parameter open-weight Llama 3 model, released April 2024.
Key Specifications
- Context window
- 8K tokens
- Knowledge cutoff
- Released
- Parameters
- 70B
- Inputs
- Text
Hosted API pricing
| Provider | Input /1M tokens | Output /1M tokens | Cost at 10M in + 2M out | |
|---|---|---|---|---|
Replicate
|
$0.65 | $2.75 | $12.00 | View |
|
Meta
Creator
|
Open weights, no hosted price | View | ||
Heads up: Base-tier, on-demand rates per 1M tokens; cached, batch and long-context tiers excluded. A provider may serve a shorter context or a quantized build than the creator's release. Verify before provisioning. More on how we price.
Estimated cost to self-host
Llama 3 70B needs about 43 GB of GPU memory at 4-bit with its full 8K context. The cheapest rental that fits is 4× RTX 3060 at about $173 a month.
That costs the same as roughly 173M tokens a month on Replicate's API. Self-hosting is more expensive below that volume.
| Precision | Memory | Cheapest fit (Full 8K context) | Cost /mo | Break-even vs API |
|---|---|---|---|---|
| 4-bitINT4 / FP4 |
43 GB
|
$173
|
173M tokens /mo
|
|
| 8-bitFP8 / INT8 |
75 GB
|
$346
|
346M tokens /mo
|
|
| 16-bitFP16 / BF16 |
145 GB
|
$979
|
979M tokens /mo
|
Estimates based on median on-demand rates for Nvidia GPUs. Memory is weights plus KV cache for one request, using FP8 KV cache where supported and FP16/BF16 otherwise. Break-even assumes a 5:1 input-to-output ratio. No guarantee of runtime support, usable performance, or that a matching quantized build exists. How we estimate costs.
Similarly priced models
The models nearest Llama 3 70B by blended rate, each at its own cheapest provider.
| Model | Blended /1M | Input /1M | Output /1M | Context | Cutoff | vs Llama 3 70B |
|---|---|---|---|---|---|---|
Qwen3.6-Plus
Alibaba Cloud
|
$0.9167 | $0.50 | $3.00 | 1M | −8% | |
GLM-5.2
Z.AI
|
$0.95 | $0.70 | $2.20 | 1M | −5% | |
Nemotron 3 Ultra 550B A55B
Nvidia
|
$0.99 | $0.66 | $2.64 | 1M | Sep 2025 | −1% |
DeepSeek R1
DeepSeek
|
$1.00 | $0.70 | $2.50 | 128K | 0% | |
|
DeepSeek R1 Turbo
DeepSeek
|
$1.00 | $0.70 | $2.50 | 64K | 0% | |
|
Llama 3 70B
This model
Meta
|
$1.00 | $0.65 | $2.75 | 8K | Dec 2023 | |
|
Qwen3-VL-235B-A22B Thinking
Alibaba Cloud
|
$1.00 | $0.40 | $4.00 | 262K | 0% | |
Kimi K2.6
Moonshot AI
|
$1.0333 | $0.60 | $3.20 | 262K | +3% | |
Grok Build 0.1
xAI
|
$1.1667 | $1.00 | $2.00 | 256K | +17% | |
Command A+
Cohere
|
$1.20 | $0.80 | $3.20 | 128K | +20% | |
|
GLM-5.1
Z.AI
|
$1.2417 | $0.83 | $3.30 | 205K | +24% |
Prices are USD per 1M tokens at each model's cheapest listed provider. Blended is the cost of 10M input plus 2M output tokens, spread over the 12M.
Frequently Asked Questions
What is Llama 3 70B good for?
Text-only generation, chat and classification on your own hardware. Open weights under the Meta Llama 3 Community License, servable on a single high-memory GPU.
When is Llama 3 70B not a good fit?
Long documents: it takes far shorter inputs than the later 70B Llamas. No image input or tool calling, and its knowledge is older than the current open models'.
What is the cheapest way to run Llama 3 70B?
Hosted, unless you push serious volume. Replicate charges $0.65 in / $2.75 out per 1M tokens. The cheapest rental that fits is 4x RTX 3060 at $173 a month, which costs the same as about 173M tokens a month on that API.
Can I self-host Llama 3 70B?
Yes. The weights are released under the Meta Llama 3 Community License. At 4-bit it needs about 43 GB of GPU memory, which starts at roughly $173 a month on the cheapest rental that fits.
More from Meta
| Model | Context | Input /1M | Output /1M |
|---|---|---|---|
|
Muse Spark 1.1
|
1M | $1.25 | $4.25 |
|
Muse Spark 1.2
|
1M | $1.25 | $4.25 |
|
Muse Spark 1.3
|
1M | $1.25 | $4.25 |
|
Muse Glimmer 30B
Open weights
|
131K | $0.20 | $0.80 |
|
Llama 4 Maverick
Open weights
|
1M | $0.20 | $0.70 |
|
Llama 4 Scout
Open weights
|
10M | $0.16 | $0.64 |
|
Llama 3.3 70B
Open weights
|
128K | $0.13 | $0.40 |
|
Llama 3.2 11B Vision
Open weights
|
128K | $0.05 | $0.68 |