
How do I sign up for together.ai and buy credits (what’s the minimum purchase) to start using the API?
Getting started with together.ai is intentionally low-friction: you create an account, receive free credits automatically, add billing when you’re ready, and then call the OpenAI-compatible API from your app or notebook. There’s no long-term commitment or large upfront minimum; you can start small, validate latency/cost, and scale up as usage grows.
Quick Answer: You sign up for together.ai by creating an account on the website, which includes free starter credits to use the API. After that, you can add a payment method and purchase pay‑as‑you‑go credits (no large minimum purchase or long-term contract required) to continue running workloads at scale.
The Quick Overview
- What It Is: A self-serve way to create a together.ai account, get an API key, and fund usage with credits so you can run open-source and partner models on the AI Native Cloud.
- Who It Is For: Developers, ML engineers, and AI product teams who want to run or migrate workloads (chat, RAG, vision, batch, fine-tuning) on an OpenAI-compatible API without managing GPU infrastructure.
- Core Problem Solved: It removes upfront friction—no GPU setup, no enterprise contract required—so you can benchmark latency and cost on real workloads using a simple, credit-based billing model.
How It Works
Once you understand the flow, going from “no account” to “tokens on the wire” takes only a few minutes. At a high level:
- Create Your together.ai Account
- Get Your API Key and Free Credits
- Add Billing and Top Up Credits as You Scale
Here’s how each phase works in practice.
1. Create Your together.ai Account
- Go to https://www.together.ai.
- Click Sign up (or Get started).
- Register using:
- Your work email, or
- A supported SSO option (if available to you).
- Verify your email address if prompted.
Once your account is created, you’re on the AI Native Cloud: you can access Together Sandbox, run models via the API, and explore the docs.
2. Get Your API Key and Use Free Credits
New accounts come with free credits to start, so you can:
- Call Serverless Inference for text, vision, code, and other modalities.
- Use Together Sandbox for quick interactive experiments.
- Explore RAG, agents, and batch workflows using the examples in the docs.
To get your key:
-
After sign-up, go to the dashboard.
-
Navigate to the API Keys section.
-
Click Create API key, name it, and store it securely.
-
Use it with the OpenAI-compatible API, for example:
export TOGETHER_API_KEY="YOUR_API_KEY" curl https://api.together.xyz/v1/chat/completions \ -H "Authorization: Bearer $TOGETHER_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo", "messages": [ {"role": "user", "content": "Hello, world!"} ] }'
Or in Python (using the Together library or OpenAI-compatible clients):
from together import Together
client = Together(api_key="YOUR_API_KEY")
resp = client.chat.completions.create(
model="meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo",
messages=[{"role": "user", "content": "Test my together.ai setup."}],
)
print(resp.choices[0].message.content)
These free credits are there specifically so you can benchmark latency, tokens/sec, and unit cost on real workloads before you commit any spend.
3. Add Billing and Buy Credits (Minimum Purchase)
When you’re ready to move beyond the free tier and into real workloads:
- Open the Billing section in your together.ai dashboard.
- Add a payment method (typically a credit card; enterprise plans can arrange invoicing via sales).
- Choose your billing model:
- Pay-as-you-go: Usage is billed against your payment method as you consume tokens.
- Prepaid credits / top-up: In some setups you can buy credits ahead of time, then burn them down with usage.
Minimum Purchase:
together.ai is designed to be low-commitment and usage-based. You don’t need a large upfront commitment or fixed contract to run workloads; you can start with small usage and scale. Exact currency minimums (e.g., smallest card charge or top-up size) can change over time and by region, so the authoritative number will always be shown in your Billing screen at checkout.
If you need a specific minimum (e.g., for procurement), it’s best to:
- Check the in-app billing panel, and
- If you need custom commercial terms, contact sales to discuss a tailored plan.
Features & Benefits Breakdown
The sign-up and credit model is built to support the full together.ai stack: Serverless Inference, Batch Inference, Dedicated Inference, GPU Clusters, and Model Shaping. Here’s how that translates into concrete benefits.
| Core Feature | What It Does | Primary Benefit |
|---|---|---|
| Free Credits on Sign-Up | Automatically credits new accounts so you can call the API immediately. | Test latency, throughput, and quality with no upfront payment. |
| OpenAI-Compatible API | Lets you swap endpoints with minimal or no code changes. | Fast migration from other providers; test price-performance fast. |
| Usage-Based Billing | Bills based on tokens processed and workloads run. | No long-term commitments; start small, then scale with demand. |
| One Account, Every Modality | Use the same key for text, image, video, code, and voice workloads. | Simplifies billing and observability across your full AI surface. |
| Path to Dedicated & Clusters | Use credits initially, then move steady workloads to dedicated endpoints. | Optimize for latency/cost with Dedicated Inference or GPU Clusters. |
Ideal Use Cases
- Best for early-stage prototyping: Because you can sign up, get free credits, and hit the API within minutes—perfect for evaluating models, building PoCs, or validating latency vs your current provider.
- Best for teams planning production migration: Because you can start with serverless and pay-as-you-go credits, then gradually shift hot paths to Dedicated Model Inference or GPU Clusters once you understand traffic patterns and SLOs.
Limitations & Considerations
- Minimum purchase specifics vary: together.ai doesn’t force a large contract, but the exact smallest charge/top-up is subject to change and may differ by region or currency. Always confirm in the Billing UI or with sales if you need a guaranteed number for procurement.
- Free credits are for evaluation, not full production: They’re designed to help you benchmark and prototype. For steady, high-throughput workloads, plan to move to Dedicated Inference or GPU Clusters with a proper budget and SLOs.
Pricing & Plans
together.ai uses a usage-based model, anchored in best price-performance rather than a rigid tiered paywall. Broadly:
- You pay for:
- Tokens generated or processed (serverless / real-time),
- Batch workloads (up to 30 billion tokens per job, at up to ~50% less cost than real-time),
- Reserved capacity for Dedicated Inference or GPU Clusters when you need guaranteed performance.
Within that, you can think about “plans” as usage patterns rather than marketing tiers:
- Starter / Self-Serve Usage: Best for individual developers or small teams needing fast experimentation with no long-term commitments. Use free credits, then light pay-as-you-go spend to iterate in Together Sandbox and via the API.
- Production / Enterprise Usage: Best for organizations with predictable or high-volume traffic that want:
- Dedicated Model Inference for steady, latency-sensitive workloads.
- Dedicated Container Inference or GPU Clusters for custom runtimes and training/fine-tuning.
- SOC 2 Type II assurances, tenant-level isolation, and custom commercial terms.
For detailed per-model pricing and any current minimum charge, check your together.ai dashboard or talk directly with the sales team.
Frequently Asked Questions
Do I need to pay anything to start using the together.ai API?
Short Answer: No, you can start with free credits.
Details: When you sign up for together.ai and create an account, you receive free credits that let you immediately run workloads through the API and Together Sandbox. This is enough to:
- Call top open-source and partner models,
- Test Serverless Inference latency and throughput,
- Experiment with RAG and agent workflows.
You only need to add a payment method and buy credits once you’re ready to exceed that free usage or move into more intensive workloads.
What’s the minimum credit purchase to keep using the API after I exhaust the free tier?
Short Answer: together.ai is pay-as-you-go with no big upfront commitment; the exact minimum charge is shown in your Billing screen and may vary.
Details: After your free credits are used, you can:
- Add a credit card (or arrange invoicing for enterprises),
- Continue with usage-based billing or prepaid credits.
There is no requirement for a large contract or high minimum spend just to access the API. The smallest allowed purchase or charge (e.g., minimum top-up) can change over time and by region, so the exact number is always governed by what you see in the Billing UI at the moment you add funds. If you need a guaranteed minimum for internal approvals, contact sales for a clear, written quote.
Summary
Signing up for together.ai and funding API usage is deliberately simple: you create an account, receive free credits to start, generate an API key, and then attach a payment method once you’re ready to scale beyond evaluation. There’s no long-term contract or large upfront minimum; you pay based on usage as you move from Serverless Inference experiments to high-throughput Batch Inference, Dedicated Model Inference, or GPU Clusters.
You get access to every modality via one OpenAI-compatible API, backed by cutting-edge systems research (FlashAttention, ATLAS, CPD, Together Kernel Collection) and production guarantees (99.9% uptime, SOC 2 Type II, tenant-level isolation, and clear data ownership: your data and models remain fully under your ownership).