Grok 4.3
xAI's latest flagship Grok model with 1M context, image input, configurable reasoning, and strong agentic tool use.
You always get the exact model you pick — we never silently route you to another.
About Grok 4.3
Grok 4.3 made a deliberate trade: xAI stopped chasing frontier performance and built something more practical instead. The result is a model that earns its keep through native X/Twitter integration — pulling posts seconds old when news breaks — and a 1 million token context window that handles entire codebases or lengthy regulatory documents in a single pass. At $1.25 per million input tokens, it arrives 40-60% cheaper than its predecessor Grok-4, and users find real value in its DeepSearch mode, which combines live web data with X discussions in a way that rivals Perplexity for current-events research. Frontend developers report genuinely polished web UI output, moving past the "cheap AI demo" look. The honest trade-off: creative writers consistently find it too literal and verbose, and its 16-second time-to-first-token sits at the high end for reasoning models in this price range. If your work is anchored in real-time information or long-document analysis rather than narrative craft, Grok 4.3 offers a focused, cost-sensible tool.
Best for
- Real-time news tracking and social media trend analysis via native X/Twitter integration
- Long-document analysis — processing full codebases, research corpora, or regulatory PDFs within a 1M token context window
- Current events research using DeepSearch mode, which combines live web results with X discussions
- Rapid frontend prototyping that produces professionally polished web UI code
- Security audits and vulnerability assessments where less restrictive guardrails are an advantage
Specifications
| Provider | xAI |
|---|---|
| Released | 2026-04 |
| Intelligence | High |
| Speed | Fast |
| Context window | 1,000,000 tokens |
| Max output | 1,000,000 tokens |
| Knowledge cutoff | December 2025 |
| Input price | $1.25 / 1M tokens |
| Output price | $2.50 / 1M tokens |
| Request cost | 1 premium request |
| Plan tier | Premium |
| Input | Text, Image |
| Output | Text |
| Features | 10M TPM · 1,800 RPM, Cached input: $0.20 / 1M tokens, Higher context pricing applies above 200K context, Configurable reasoning: none, low, medium, or high; defaults to low, Structured outputs, web search, X search, function calling, and code execution supported |
| API endpoints | v1/chat/completions, v1/responses, v1/batch |
| Model ID | grok-4.3 |
Frequently asked questions
Standard pricing is $1.25 per million input tokens and $2.50 per million output tokens. Cached inputs drop to $0.20 per million, which can cut effective input costs by up to 90% on repeated context. Requests exceeding 200k input tokens are billed at $2.50 per million input tokens.
Grok 4.3 supports a 1 million token context window, with up to 1 million tokens per response — suitable for processing entire codebases or lengthy documents in a single pass.
Its native X/Twitter integration is genuinely unique: it can pull posts that are seconds old, making it the strongest option for breaking news tracking, trend analysis, and social media monitoring. No other major model has this at the same depth.
Creative writing is a consistent weakness — users describe it as too literal and robotic with weak narrative coherence. It also tends toward verbose output, and its 16-second time-to-first-token is high relative to competing models in its price range.
Grok 4.3 is 40-60% cheaper than Grok-4 (which launched at $3/$15 per million input/output), adds native video input, and includes architectural improvements. However, it does not significantly surpass Grok-4 in raw capability and is positioned as a practical productivity tool rather than a frontier model.
Journalists, social media managers, researchers, and teams doing current-events work will benefit most from the X integration and DeepSearch mode. It is also a strong pick for cost-sensitive applications needing a large context window, or developers prototyping web UIs. Writers and users needing fast response times should look elsewhere.