xAI
Updated May 2026
Grok
Elon Musk's X-integrated chatbot with a sharper tongue
Founded by Elon Musk in 2023, xAI positions Grok as a 'maximum truth-seeking' AI that's willing to answer spicy questions other models refuse. It's trained with real-time data from X (formerly Twitter) and runs on Colossus, one of the largest GPU clusters in the world. The brand leans into rebellious, irreverent personality but the underlying models are genuinely competitive with OpenAI and Anthropic on reasoning benchmarks.
Variants
6
Best at
2M token context on Grok 4.1 Fast
Max context
2M
tokens
Pricing
Free
$0
per X users with daily limits
SuperGrok
$30
per month
SuperGrok Heavy
$300
per month, includes 4.3 Beta
Grok 4 API
$3 in / $15 out
per million tokens
Grok 4.1 Fast API
$0.20 in / $0.50 out
per million tokens
Variants
Sort the table by context window or cost to find the right variant. Click any version below for a battle card with ranks, pricing notes, and official links.
| Modalities | ||||
|---|---|---|---|---|
Grok 4.1 Fast grok-4-1-fast | 2M | varies / varies | 2025 | text |
Grok 4.3 Beta grok-4-3 | 256K | varies / varies | 2026 | textvisionvideoaudio |
Grok 4.1 grok-4-1 | 256K | varies / varies | 2025 | textvision |
Grok 4 grok-4 | 256K | varies / varies | 2025 | textvision |
Grok Code Fast grok-code-fast | 256K | varies / varies | 2025 | code |
Grok 3 grok-3 | 131K | $3.00 / $15 | 2025 | textvision |
Battle card
Context rank
#2
within Grok
Capability rank
#1
modalities + reasoning
Weights
Closed
API/app access
Best fights to pick
- generating ready-to-send business documents
- video-based workflows
- X power users
Rankings are Tendril directory ranks, computed from the model data shown here. Public benchmark leaderboards change often, so official docs and current benchmark pages should be checked before buying or deploying.
Learn
Lessons about this model
Structured lessons that cover Grok directly or put it in context alongside its rivals.
Grok 4.1 Fast — when 2M context beats a smarter model
Grok-Code — coding benchmarks and reality
Grok Vision — visual reasoning on the third option
Grok — When X's Firehose Matters
Claude Code vs. Codex CLI vs. Grok Code — the coding agent picker
Claude vs. ChatGPT vs. Gemini — Side-by-Side
Check yourself
Quizzes
Short, mixed-difficulty quiz sets on Grok and its model family.
Hands-on
Try these prompts
Ready-made prompts that show Grok at its best. Use them in your own AI workspace, then compare the output with what you learned in Tendril.
Real-time X firehose question
BuildersGrok's edge is live X data — ask for something that would be stale on any other model.
What are people on X saying about this week's tech layoffs? Summarize the top themes and cite 3 representative posts with handles.
Grok for less-filtered commentary
BuildersGrok tends to be more willing to take a side on contested topics — compare its answer against Claude or GPT.
Give me your actual opinion, not a 'both sides' summary: is remote work making software engineering better or worse overall? One paragraph, take a position, defend it.
Grok Code on a benchmark snippet
BuildersFeed it a LeetCode-style problem and compare latency and quality with Claude/GPT.
Write a Python function that returns all pairs in an array summing to target, in O(n) time. Include tests. Explain the tradeoff vs. the naive O(n^2) version in two sentences.
Print & keep
Printable reference
One-page summaries and flowcharts — great for desks, classrooms, or study sessions.
Go deeper
Official resources
Straight from the lab — docs, API references, and the chat surfaces you can try today.
Strengths
- 2M token context on Grok 4.1 Fast
- willing to engage with controversial topics
- deep X/Twitter data access
Limits
- knowledge cutoff November 2024 on 3/4
- smaller ecosystem than OpenAI or Google
- political framing turns off some users