What other models does xAI offer?

xAI offers a range of Grok models from lightweight general-purpose options to frontier reasoning variants with extended context. Here are some other models xAI supports:

• Grok 4.20: Grok 4.20 is a reasoning model variant in the Grok 4 family priced at $2.00 per million input tokens and $6.00 per million output tokens, offering a 2 million token context window that doubles Grok 4.3's capacity for workloads requiring very large input processing

• Grok 4: Grok 4 is xAI's flagship reasoning model and the base of the Grok 4 generation, available through the xAI API and to SuperGrok subscribers, establishing the reasoning performance ceiling that Grok 4.3 builds on with improved production pricing

• Grok 4 Heavy: Grok 4 Heavy is xAI's multi-agent reasoning model, designed for the most demanding tasks with parallel test-time compute and a 256,000 token context window, restricted to the SuperGrok Heavy subscription tier

• Grok 3: Grok 3 is a non-reasoning general-purpose model priced at $4.00 per million input tokens, suited for instruction following and text generation workloads where extended chain-of-thought processing is not required

• Grok 3 Mini: Grok 3 Mini is the smallest offering in the Grok 3 family, targeting high-volume or latency-sensitive applications where cost efficiency takes priority over reasoning capability

How does Grok 4.3 differ from xAI's other models?

Grok 4.3 is positioned as a production-ready reasoning model in the Grok 4 family, released April 30, 2026, balancing high intelligence with a significantly lower per-token cost compared to earlier Grok 4 variants.

• Pricing: Grok 4.3 costs $1.25 per million input tokens and $2.50 per million output tokens, with cache hits at $0.20 per million tokens (an 84% cache discount). This makes it substantially cheaper than Grok 4.20 ($6.00/M output) and base Grok 4, while delivering reasoning capability that non-reasoning models like Grok 3 cannot match

• Context window: Grok 4.3 supports a 1 million token context window, the same as Grok 4 and half of Grok 4.20's 2 million token limit, which is sufficient for the vast majority of long-document and multi-turn reasoning tasks

• Intelligence Index: Grok 4.3 scores 53 on the Artificial Analysis Intelligence Index, ranking #13 of 150 evaluated models, placing it above Grok 4.20 (49) in the current rankings and making it one of the highest-scoring models across all providers

• Speed: At 140.6 output tokens per second, Grok 4.3 is slower than Grok 4.20 (201.5 t/s) but its TTFT of 16.42 seconds reflects the extended reasoning processing that drives its high intelligence score

• Verbosity: Grok 4.3 generated 88 million tokens during evaluation, above the 35 million median for comparable models, indicating highly detailed responses that can increase output token costs per request without prompt-level length controls

• Use case fit: Grok 4.3 is the best option for production reasoning workloads where intelligence ranking matters most and cost efficiency is a primary constraint alongside capability

Grok 4.3 is the practical choice for teams that need top-tier reasoning performance without paying the higher output token rate of Grok 4.20.

What models should I consider using alongside Grok 4.3?

No single model is optimal for every task. Here are models worth pairing with Grok 4.3 depending on what your product needs:

• Grok 4.20 (xAI): When a specific request requires processing more than 1 million tokens of context, routing to Grok 4.20 provides the 2 million token window while staying within xAI's provider ecosystem, with no API credential changes required

• Claude Sonnet 4 (Anthropic): For structured output generation, document transformation, or tasks where response formatting precision is critical, Claude Sonnet 4 provides a well-benchmarked alternative with strong instruction-following characteristics and a different provider risk profile

• Gemini 2.5 Flash (Google): For high-volume, low-complexity requests in the same pipeline where Grok 4.3's reasoning depth is unnecessary, Gemini 2.5 Flash offers fast responses and competitive pricing to reduce pipeline cost on simpler tasks

• DeepSeek R1 (DeepSeek): For cost-sensitive reasoning workloads where provider diversity is a goal, DeepSeek R1 provides chain-of-thought capability at a significantly lower price point and serves as a strong cross-provider fallback

• GPT-5 (OpenAI): For applications requiring provider redundancy at the frontier reasoning tier, GPT-5 delivers comparable intelligence from a separate infrastructure, reducing the risk of a single-provider outage affecting high-stakes reasoning requests

What are the challenges of using Grok 4.3 in my product?

Like any production LLM, Grok 4.3 comes with tradeoffs worth planning for:

• High TTFT for latency-sensitive interfaces: A time-to-first-token of 16.42 seconds is appropriate for batch reasoning tasks but unsuitable for user-facing interfaces expecting near-instant response starts. Applications requiring low-latency first tokens need a non-reasoning fallback or streaming with careful UX design around wait time

• Verbosity driving output cost: Grok 4.3 generated 88 million tokens per evaluation run versus a 35 million median for comparable models. Without explicit length constraints in your prompts, responses trend long, raising output token costs that can compound quickly at production request volumes

• Provider dependency: All traffic to Grok 4.3 flows through xAI's infrastructure, with no major cloud marketplace redundancy. A rate limit change, API policy update, or infrastructure incident at xAI directly impacts availability unless you have cross-provider failover configured

• Cache discount requires prompt engineering: The 84% cache hit discount is compelling at $0.20 per million tokens, but capturing it requires consistent prompt prefixes and deliberate caching strategy. Teams that haven't invested in prompt cache architecture will pay the full $1.25 input rate

• No multimodal output: Grok 4.3 accepts both text and image inputs but produces text-only output. Workflows requiring generated images or other output modalities need routing to a separate model

Why should I use Merge Gateway to route LLM requests with Grok 4.3 and every other model?

Using Grok 4.3 through Merge Gateway gives you access to the model itself and the infrastructure layer around it:

• One API, every provider: Route to Grok 4.3 or any other major LLM through a single endpoint and API key. Swapping providers requires only a model string change and no modifications to application code

• Intelligent routing and automatic failover: Merge routes around xAI outages automatically. Routing policies based on cost, latency, or quality can reduce spend by 40–60% without touching your application code

• Cost governance: Set hard or soft project budgets so Grok 4.3 spend stays within plan. Every request is attributed to a model, project, and tag in a unified billing dashboard across all providers

• Build Your Own Router: Define what "best" means for your traffic by selecting from curated ML benchmarks or adding your own eval scores. The router scores each available model against your weights and picks the winner per request, with a plain-language explanation of every decision

• Security and compliance controls: Apply DLP rules and prompt injection protection before every request reaches xAI. Enforce per-project model and region policies without adding that logic to your application

How can I start routing requests to Grok 4.3 via Merge Gateway?

Getting Grok 4.3 running through Merge Gateway takes a few minutes:

1. Create an account and get your API key from the dashboard.

2. Install the Merge Gateway SDK: run pip install merge-gateway-sdk (Python) or npm install merge-gateway-sdk (Node). Alternatively, if you're already using the OpenAI SDK, set base_url = "https://api-gateway.merge.dev/v1/openai" and your existing code works as-is.

3. Make your first request using the provider/model format. For Grok 4.3, the model string is xai/grok-4.3. Swap the model string to route to any other provider without changing anything else.

4. Configure a routing policy in the dashboard to set failover behavior, cost limits, and optimization strategy. Your first policy can be as simple as naming Grok 4.3 as primary for reasoning tasks with a lower-cost model as a fallback for simpler classification or summarization requests.

Full setup instructions and SDK references are in the Merge Gateway docs.