TL;DR: Grok is xAI's model family, and in mid-2026 it is small and cheap. Grok 4.3 is the flagship at $1.25 input and $2.50 output per million tokens, with a 1M context window. Grok 4.1 Fast is the budget tier at $0.20/$0.50 with a 2M window. There are also coding, image, and video models, plus live access to data from X.
xAI keeps Grok simple. After a cleanup in May 2026, the old Grok 3 and Grok 4 versions were retired, and almost everything now runs on Grok 4.3, the flagship. It is fast, smart, and flat-priced at $1.25 input and $2.50 output per million tokens.
Around it sit a few purpose-built models: the Grok 4.20 agentic variants for long, multi-step work, Grok 4.1 Fast for cheap high-volume tasks, grok-build for coding, and grok-imagine for images and video.
Key takeaways
- Grok 4.3 is the flagship, flat-priced at $1.25/$2.50 per million tokens with a 1M context window.
- Grok 4.1 Fast is the budget tier at $0.20/$0.50 and a huge 2M context window.
- The agentic Grok 4.20 variants add reasoning, non-reasoning, and multi-agent modes for long tasks.
- Grok can read live data from X, which suits real-time search, news, and social apps.
- xAI retired Grok 3, Grok 4, and the old fast models in May 2026. They now redirect to Grok 4.3.
The current Grok models at a glance
Here is the lineup, with the model name your engineers will call. Use this table as your reference. Everything below explains the details. Prices are input then output, per one million tokens.
| Model | Context | Price in / out (per 1M) | Best for |
|---|---|---|---|
| Grok 4.3 | 1M | $1.25 / $2.50 | Flagship, fast and most intelligent |
| Grok 4.20 (reasoning) | 2M | $1.25 / $2.50 | Long, hard, step-by-step tasks |
| Grok 4.20 (multi-agent) | 2M | $1.25 / $2.50 | Agent teams working in parallel |
| Grok 4.1 Fast | 2M | $0.20 / $0.50 | Cheap, fast, high-volume work |
| grok-build-0.1 | 256K | $1.00 / $2.00 | Agentic coding sessions |
| grok-imagine | — | per image / second | Image and video generation |
Which Grok model fits your team?
Pick your main use case to see the right model.
For chatbots, tagging, and routing at scale, Grok 4.1 Fast is cheap at $0.20/$0.50 per million tokens and carries a 2M context window. You only pay for the flagship when a task needs it. See what building this in-house costs on our developer rate card.
For coding and long agent runs, grok-build is tuned for code and the Grok 4.20 multi-agent variant runs agent teams in parallel. Pair them with engineers who ship agents. Hire AI agent developers who build with them.
For news, search, and social apps that need fresh data, Grok can read live from X, which most rivals cannot. Strong AI and machine learning engineers get the most from it.
The model is the easy part. The hard part is the team that builds and ships with it. We match you with pre-vetted engineers in about 24 hours. Tell us what you need.
The Grok lineup in plain English
xAI splits Grok into a flagship, a fast tier, agentic variants, and a set of media models. The flagship does most of the work. The others are for specific jobs. Once you see the shape, the list is short and easy to hold in your head.
Grok 4.3 and the 4.20 variants
Grok 4.3 is the default. xAI calls it the most intelligent and fastest model, and it is flat-priced at $1.25 input and $2.50 output per million tokens with a 1M context window. The Grok 4.20 line adds three agentic variants: a reasoning mode that thinks step by step, a non-reasoning mode for fast answers, and a multi-agent mode that runs several agents in parallel. These carry a 2M context window for long, multi-step work.
Grok 4.1 Fast, grok-build, and grok-imagine
Grok 4.1 Fast is the budget tier at $0.20 input and $0.50 output, with a 2M context window and prompt caching that drops cached input to $0.05. It is the model for high-volume, simple work. grok-build is the coding model, tuned for agentic code sessions with a 256K window. grok-imagine generates images and video. Together they cover text, code, and media from one platform and one API key.
What makes Grok different
The standout feature is live data. Grok can search X in real time and fold fresh posts into an answer. Most closed models only know their training data up to a cutoff. For news, trends, and social listening, that live feed is a real edge. Grok also leans into agents, with a multi-agent mode built into the 4.20 line rather than bolted on.
The other headline is context. A 2M token window on the Fast and agentic models is about 1.5 million words. That is enough for very large codebases, long document sets, or extended agent runs that keep a lot of history. Every current Grok model also reads images and calls tools, so you can build agents that see and act, not just chat.
Pricing compared across the lineup
Grok is cheap for its class. Output tokens drive most cost, and the flagship output is $2.50 per million, well below the flagship price of most Western labs. Grok 4.1 Fast output is $0.50, five times cheaper again. The chart below shows output price across the main models.
The best way to save is to route by difficulty. Send simple, high-volume calls to Grok 4.1 Fast and keep only the hard ones on Grok 4.3 or the agentic variants. Prompt caching cuts repeated context to a fraction of the input price. Teams that build this routing well often cut their bill by half. This is the kind of work that separates a strong AI engineer from an average one.
Context windows and reasoning modes
The context window is how much text the model reads at once, counting your prompt and the conversation so far. Grok 4.3 offers a 1M window, roughly 750,000 words. The Fast and agentic 4.20 models double that to 2M. grok-build uses a 256K window, which is plenty for most coding sessions. Larger windows let you drop whole codebases or document stacks into one request.
The 4.20 line splits reasoning from non-reasoning. In reasoning mode the model thinks step by step before it answers, which helps on hard math, code, and logic but uses more tokens. In non-reasoning mode it answers fast, which suits chat and simple tasks. Your engineers should pick the mode on purpose. An engineer who tunes this well runs Grok far cheaper at the same quality. You can read more in our breakdown of Claude versus ChatGPT for coding.
Which model should you use
Start with Grok 4.3. It is the safe default for most tasks. Move down to Grok 4.1 Fast when the task is simple and you run high volume. Move up to the agentic 4.20 variants for long, multi-step work. Use grok-build for coding and grok-imagine for media. The table below maps common jobs to a model.
| Job | Recommended model | Why |
|---|---|---|
| Support chatbot, FAQ | Grok 4.1 Fast | Fast and cheap at high volume |
| Data tagging, routing | Grok 4.1 Fast | Simple, repeated, low risk |
| Daily coding and code review | grok-build-0.1 | Tuned for agentic code sessions |
| Real-time search, news, social | Grok 4.3 | Live X data, fresh answers |
| Long agent runs, multi-agent | Grok 4.20 multi-agent | Agent teams, 2M context |
| Images and video | grok-imagine | Built-in media generation |
A simple rule helps. If a cheaper model gives the right answer most of the time, use it and add a check. Escalate only the cases that fail. This keeps quality high and cost low. We worked with a media team that ran everything on the flagship. Moving routine tagging and search to Grok 4.1 Fast cut their bill by more than half with no drop in quality.
Legacy and retired models
xAI moves fast and retires old models. In May 2026 it retired Grok 3, Grok 4, Grok 4 Fast, Grok 4.1, Grok 4.1 Fast’s first release, and the old code model. Calls to those IDs now redirect to Grok 4.3 and bill at the flat rate. The table shows the main older models and what to use instead.
| Older model | Status | Move to |
|---|---|---|
| Grok 4, Grok 4 Fast | Retired, redirects | Grok 4.3, Grok 4.1 Fast |
| Grok 3, Grok 3 mini | Retired | Grok 4.3 |
| grok-code-fast-1 | Retired | grok-build-0.1 |
| Grok 2, Grok 1 | Legacy, open weights | Grok 4.3 for new work |
A move is usually small. Change the model name and check that your context and pricing assumptions still hold, since the newer models often have larger windows. The xAI models page lists every current and retired ID in one place, so check it before you migrate.
What this means for hiring
Knowing the lineup is step one. Getting value from it is the harder step. With Grok the biggest levers are routing between the flagship and the Fast tier, using live X data well, and building clean multi-agent flows. The teams that win are the ones who do this, not the ones on the biggest model. That is engineering, not just a model choice.
Strong AI engineers are scarce and costly in the United States. A senior AI engineer there runs $12,000 to $18,000 per month all in. The same skill from a pre-vetted engineer in Vietnam or the Philippines often costs $3,000 to $6,000 per month. The work is the same. The bill is not. Many of our clients build their whole Grok stack with engineers from our network.
If you want to go deeper on the wider picture, our library of articles on AI tools and trends covers adoption, cost, and how the labs compare. Read that next to see how fast this market is moving and why the right team matters more each quarter.
Compare other model families
Build your Grok stack with the right team
Grok gives you a cheap flagship, live data, and strong agent tools. The model is the easy part. The team that builds with it is what ships value. We match you with pre-vetted senior engineers across Asia in about 24 hours, with no upfront cost and payroll handled.
