MT
Meta Llama Models
Meta builds the open-weight Llama family, a popular base for cost-sensitive and self-hosted workloads. On thalam you can call 4 Meta Llama models through one OpenAI-compatible API, pay per token with no Meta Llama account and no subscription.
base_url = https://api.thalam.ai/v1
api_key = tl-live-... # from your thalam dashboard
model = meta-llama/llama-3.3-70b-instruct # any model id below
Meta Llama Text models
| Model | Context | Price | Speed |
|---|---|---|---|
| Llama 3.3 70B Instruct | 128K | $0.130 / 1M | Medium |
| Llama 4 Maverick | 128K | $0.270 / 1M | Fast |
| Llama 4 Scout | 128K | $0.180 / 1M | Fast |
| Llama 3.1 8B Instruct | 128K | $0.020 / 1M | Fast |
Call Meta Llama models through one API
One OpenAI-compatible endpoint, one bill. Pay per use, no Meta Llama account, credits never expire.
Meta Llama models in our best-of lists
FAQ
Do I need a Meta Llama account to use Meta Llama models?
No. thalam gives you Meta Llama's models through one API key, billed per use, so you do not need a separate Meta Llama account or key.
How much do Meta Llama models cost on thalam?
You pay per token (or per image, video or second) at the rate shown on each model page, with no platform fee and no subscription.
Which Meta Llama model should I use?
It depends on your task. See the table above for pricing and specs, or the Best AI Models lists for task-by-task picks.