Resposta curta: as of July 23, 2026, neither original API model is current. xAI retired Grok 4.1 Fast on May 15, 2026 and routes its old reasoning and non-reasoning slugs to Grok 4.3. Google shut down Gemini 3 Pro Preview on March 9, 2026 and tells developers to migrate to Gemini 3.1 Pro Preview.
For a new project, the more useful comparison is Grok 4.5 vs Gemini 3.1 Pro. Gemini publishes the broader mixed-media input set and a 1,048,576-token input limit. Grok 4.5 publishes a 500,000-token context, lower output-token pricing for prompts below 200,000 tokens, and first-party Web Search and X Search tools. The right choice depends on your files, tools, latency target, and cost per accepted result.
GlobalGPT brings multiple model routes into one place, including Grok 4.1 Fast, Grok 4.3, Gemini 3 Pro, and Gemini 3.1 Pro labels on its current model list. GlobalGPT is a third-party platform, so its model labels, limits, and billing are its own service terms; they do not keep a retired xAI or Google API endpoint active.
Official model status, checked July 23, 2026: calls to
grok-4-1-fast-reasoningare now served bygrok-4.3with low reasoning effort, while calls togrok-4-1-fast-non-reasoningare served bygrok-4.3with no reasoning effort. Google listsgemini-3-pro-pré-visualizaçãoas shut down and directs developers to change togemini-3.1-pro-preview. New integrations should use an explicit current model ID and rerun quality, latency, tool-use, and cost tests.
Comparing the Specs: Grok 4.1 vs Gemini 3 Pro Today
Plataforma de IA completa para escrita, geração de imagens e vídeos com GPT-5, Nano Banana e muito mais
The original model names remain useful for migration searches, but they no longer describe two active official API endpoints. The table below separates the retired models from the current models a developer can intentionally select.
| Especificação | Current xAI route | Current Google route |
|---|---|---|
| Legacy status | Grok 4.1 Fast retired May 15, 2026; old slugs redirect to Grok 4.3 | Gemini 3 Pro Preview shut down March 9, 2026; Google requires migration |
| Current model compared | grok-4.5 | gemini-3.1-pro-preview |
| Published context or token limit | 500,000-token context | 1,048,576 input tokens; 65,536 output tokens |
| Published input scope | xAI’s Grok 4.5 overview centers on text, tools, code, and agentic work; media generation and voice use separate xAI APIs | Text, image, video, audio, and PDF input; text output |
| Raciocínio | Low, medium, or high; high is the default | Thinking supported |
| Major tools | Function calling, Web Search, X Search, code execution | Function calling, code execution, File Search in AI Studio, Google Search, Google Maps, URL context, structured outputs |
| API rate below 200K prompt tokens | $2 input, $0.30 cached input, $6 output per 1M tokens | $2 input, $0.20 cached input, $12 output per 1M tokens |
| API rate at or above 200K prompt tokens | $4 input, $0.60 cached input, $12 output per 1M tokens | $4 input, $0.40 cached input, $18 output per 1M tokens |
xAI specifically recommends Grok 4.3 as the migration target for retired Grok 4.1 Fast workloads, while Grok 4.5 is its current flagship model for code and broader agentic work. Google directly recommends Gemini 3.1 Pro Preview as the replacement for Gemini 3 Pro Preview. Because one replacement is a preview model and both providers can change routing, pin the exact model ID returned in your logs.
Gemini has the clearer published advantage for mixed-media input and long prompts. Grok has the lower listed output rate below 200,000 prompt tokens and a dedicated X Search tool. Neither advantage proves higher answer quality for every task.
Multimodal Capabilities: Current Successors
Gemini 3.1 Pro Preview’s Published Multimodal Scope
- Inputs: text, images, video, audio, and PDF; the published output type is text.
- Token limits: 1,048,576 input tokens and 65,536 output tokens.
- Document tools: code execution, URL context, structured output, Google grounding, and File Search in AI Studio.
Those published inputs make Gemini 3.1 Pro Preview the safer documented choice when one request must compare a report, charts, screenshots, audio, and video. A large context window increases capacity, not reliability, so important claims still need citations and output checks.
Grok 4.5’s Documented Scope
- Core model: xAI positions Grok 4.5 for code, chat, tool calling, agentic tasks, and knowledge work.
- Contexto: the current xAI model catalog lists 500,000 tokens.
- Separate media routes: xAI documents image and video generation through Imagine APIs and voice through the Voice API.
The original screenshot demonstrates one response, not a reproducible benchmark. Compare both current models with the same documents, tool permissions, prompt, scoring rubric, and number of trials before selecting a production route.
Ecosystem Integration: X Search vs Google Grounding
Grok with X Search
xAI’s documentation states that Grok does not know real-time events unless search tools are enabled. Grok 4.5 supports Web Search and X Search, and xAI’s current pricing page lists each tool at $5 per 1,000 calls in USD. X Search can search posts, profiles, and threads; it is a paid tool call, not automatic live knowledge inside every answer.
Gemini with Google Grounding and File Tools
Gemini 3.1 Pro Preview supports Google Search grounding, Google Maps grounding, URL context, code execution, and File Search in AI Studio. On Google’s paid API tier, the first 5,000 grounded prompts per month are listed at no charge across Gemini 3 models, followed by $14 per 1,000 search queries. One prompt can create multiple queries, so the bill should be measured from actual logs.
Qual você deve escolher?
- Choose Grok with X Search when X posts, profiles, and threads are first-class evidence.
- Escolha Gemini when mixed-media files, web grounding, Maps, URLs, and Google-connected research are central.
- For high-stakes work, record citations, timestamps, tool calls, and unsupported claims before using the result.
Performance Benchmarks: How to Compare Current Models
Old leaderboard positions and one-off response-time tests no longer decide this comparison. The original endpoints are not current, and scores change with the model version, reasoning level, prompt, tool access, sampling settings, and evaluation date.
- Pin
grok-4.5egemini-3.1-pro-preview, and log the returned model and date. - Give both models the same tasks, files, tools, time budget, and acceptance criteria.
- Separate closed-book reasoning tests from grounded search tests.
- Run enough trials to capture variability instead of keeping one favorable response.
- Measure correctness, citation support, latency, review time, token use, tool calls, and total cost per accepted result.
For a wider shortlist, compare maintained options in GlobalGPT’s guide to the best AI models. The practical winner is the model that meets your own acceptance test at the lowest review-adjusted cost.
Developer Tools: Grok vs Gemini for Coding
Grok 4.5 for Coding and Agentic Work
xAI describes Grok 4.5 as its flagship model for code and broader agentic tasks. It supports configurable reasoning, function calling, search, and code execution through the Responses and Chat Completions APIs. This makes it a strong candidate for iterative code repair and tool-driven workflows.
Gemini 3.1 Pro Preview for Software Engineering
Google positions Gemini 3.1 Pro Preview for software engineering, precise tool use, and multi-step execution. It also offers gemini-3.1-pro-preview-customtools for workflows that mix bash and custom tools, while warning that quality can fluctuate on tasks that do not benefit from those tools. For related tooling context, see the Google Antigravity guide.
Which Should Developers Choose?
- Start with Grok 4.5 for coding-centered agent loops, X-aware tools, and lower short-context output pricing.
- Start with Gemini 3.1 Pro Preview for repositories combined with PDFs, images, video, audio, URLs, or Google grounding.
- Ship only after testing: generated screenshots and plausible code are not substitutes for passing tests and human review.
User Experience: Conversational vs Structured Responses
Grok is commonly chosen for a more conversational style, while Gemini is often chosen for structured research and business-oriented output. Tone is not a fixed technical capability: system instructions, reasoning settings, safety policies, and product interfaces can change how either model responds.
- Creative and social workflows: test whether Grok’s style reduces editing time without reducing factual support.
- Research and enterprise workflows: test whether Gemini’s structure improves review, citations, and reuse.
- Customer-facing use: apply the same safety, escalation, privacy, and quality controls to both models.
Pricing Plans: Verified Current Costs
As of July 23, 2026, xAI and Google publish the following usage-based rates in USD per one million tokens. These are API prices, not monthly subscriptions. Neither developer pricing page states that tax is included; applicable tax and currency conversion depend on the customer’s billing account and jurisdiction.
| API route | Prompt size | Entrada | Entrada em cache | Saída | Condição importante |
|---|---|---|---|---|---|
| Retired Grok 4.1 Fast slug, now served by Grok 4.3 | Below 200K | $1.25 | $0.20 | $2.50 | Old reasoning slug gets low reasoning; old non-reasoning slug gets no reasoning |
| Retired Grok 4.1 Fast slug, now served by Grok 4.3 | 200K or more | $2.50 | $0.40 | $5.00 | xAI applies long-context rates to all tokens once the prompt reaches 200K |
| Grok 4.5 | Below 200K | $2.00 | $0.30 | $6.00 | Search and code-execution tools are billed separately |
| Grok 4.5 | 200K or more | $4.00 | $0.60 | $12.00 | 500K total context |
| Visualização do Gemini 3.1 Pro | Up to 200K | $2.00 | $0.20 | $12.00 | Output price includes thinking tokens; no free API tier is listed |
| Visualização do Gemini 3.1 Pro | Over 200K | $4.00 | $0.40 | $18.00 | Context-cache storage is listed separately at $4.50 per 1M tokens per hour |
Grok 4.5 has the lower output rate below the long-context threshold. Gemini costs more for output, but its published one-million-token input limit and mixed-media support may reduce the need for a separate file pipeline. Compare total cost per accepted result, including tool calls, retries, and human review.
Consumer Subscriptions Are Separate
On Google’s United States Gemini subscription page, Free is $0 per month, Google AI Plus is $4.99 per month, Google AI Pro is $19.99 per month, and Google AI Ultra starts at $99.99 per month for 5x Pro usage or $199.99 per month for 20x. The page presents monthly billing, and access varies by plan, country, age, and language. The page does not state whether the displayed US dollar prices include tax.
Grok consumer access is billed separately from the xAI API. xAI’s public developer pricing page gives exact API and tool rates but does not publish one universal Grok.com monthly subscription price for every account and region. The historical consumer pricing image above should not be used to quote a current plan.
Current GlobalGPT Pricing and Features
As of July 23, 2026, the Página de pedidos do GlobalGPT lists Basic at $11.90 month to month or $5.80 per month billed annually, Pro at $19.90 month to month or $10.80 per month billed annually, and Unlimited at $49.90 month to month or $25.00 per month billed annually. The annual totals are $69.60, $129.60, and $300.00. The page uses a dollar sign but does not state the ISO currency code or whether tax is included; discounted monthly equivalents require payment for a full year.
- Multiple model routes: the current list includes Grok 4.1 Fast, Grok 4.3, Gemini 3 Pro, Gemini 3.1 Pro, and other model families.
- One account: compare chat, search, image, video, and research tools without maintaining a separate consumer subscription for every provider.
- Third-party terms: GlobalGPT controls its own routing, limits, plan allowances, and billing; provider API documentation remains the source for official endpoint status.
Considerações finais
Do not start a new official API integration on Grok 4.1 Fast or Gemini 3 Pro Preview. Use Grok 4.3 when you need the explicit xAI migration target, or evaluate Grok 4.5 as xAI’s current flagship. Use Gemini 3.1 Pro Preview when native mixed-media input, a published 1,048,576-token input limit, and Google grounding are decisive.
Choose Grok 4.5 for coding-centered agentic workflows, lower short-context output pricing, or X Search. Choose Gemini 3.1 Pro Preview for long mixed-media inputs and Google tools. If you want to compare provider routes from one interface, GlobalGPT provides a multi-model workflow, with its own limits and pricing.
FAQ: Grok 4.1 vs Gemini 3 Pro
Is Grok 4.1 still available through the xAI API?
Not as the original model. xAI retired the Grok 4.1 Fast reasoning and non-reasoning API models on May 15, 2026. The old slugs now route to Grok 4.3 with low or no reasoning effort and are billed at Grok 4.3 rates.
Is Gemini 3 Pro still available through the Gemini API?
No. Google lists Gemini 3 Pro Preview as shut down on March 9, 2026 and tells developers to migrate to Gemini 3.1 Pro Preview. Google’s page does not promise that the old model ID will continue working as an alias.
Which current model should a new project use?
Evaluate Grok 4.5 and Gemini 3.1 Pro Preview with explicit model IDs. Grok 4.5 fits coding, agentic, and X-search workflows; Gemini 3.1 Pro Preview fits large mixed-media inputs and Google-grounded research.
Which model has the larger current context window?
Gemini 3.1 Pro Preview publishes a 1,048,576-token input limit and a 65,536-token output limit. xAI lists a 500,000-token context for Grok 4.5 and a one-million-token context for Grok 4.3.
How much do the current replacement APIs cost?
Below 200K prompt tokens, Grok 4.5 is $2 input and $6 output per 1M tokens; Gemini 3.1 Pro Preview is $2 input and $12 output. At the long-context tier, Grok is $4/$12 and Gemini is $4/$18 in USD.
Does Grok include real-time X data automatically?
No. xAI says Grok has no access to real-time events unless search tools are enabled. Web Search and X Search are optional tools, each listed at $5 per 1,000 calls.
Can GlobalGPT show both legacy labels and current successors?
Yes. GlobalGPT’s current list includes Grok 4.1 Fast, Grok 4.3, Gemini 3 Pro, and Gemini 3.1 Pro. These are GlobalGPT routes under its own terms and do not change official xAI or Google endpoint retirements.
Which is better in 2026: Grok or Gemini?
Grok 4.5 is the stronger starting point for X-search and coding-centered agentic work. Gemini 3.1 Pro Preview is the stronger documented option for long mixed-media inputs and Google tools. Test both on your own acceptance criteria.
