We use cookies. Cookies and browser storage keep you signed in and remember your preferences. With your permission we also measure how the site is used. You can any time. See our cookie policy.

New here? See how it works A 1-minute tour of the playground, models and your API key.

Relace Search vs Grok 4.20: API price, context & features · api14

Relace: Relace Search vs SpaceXAI: Grok 4.20

A side-by-side comparison of Relace: Relace Search ($3.50 / 1M tokens) and SpaceXAI: Grok 4.20 ($4.38 / 1M tokens) using current USD prices and capabilities from the api14 catalog. Both run through the same OpenAI-compatible API.

Specifications

	Relace: Relace Search	SpaceXAI: Grok 4.20
Provider	Relace	xAI
Input	$3.50 / 1M tokens	$4.38 / 1M tokens
Output	$10.50 / 1M tokens	$8.75 / 1M tokens
Cache read	-	$0.70 / 1M tokens
Web search	-	$17.50 / 1K searches
Context window	256,000	2,000,000
Max output	128,000	1,800,000
Input	text	text, image, file
Output	text	text
Tool calling
Structured output
Reasoning control
Added	Dec 8, 2025	Mar 31, 2026

Key differences

Relace: Relace Search is 20% cheaper on input tokens.
SpaceXAI: Grok 4.20 is 17% cheaper on output tokens.
SpaceXAI: Grok 4.20 reads 7.8× more context (2,000,000 tokens).
Only SpaceXAI: Grok 4.20 supports structured JSON output.
Only SpaceXAI: Grok 4.20 supports reasoning effort control.
A request with 1,000 input and 500 output tokens costs $0.00875 with Relace: Relace Search and $0.00875 with SpaceXAI: Grok 4.20.

In depth

Relace: Relace Search is a text model from Relace, released on 2025-12-08. SpaceXAI: Grok 4.20 is a text model from xAI, released on 2026-03-31. Both are served as text-in, text-out chat models on api14, so from a calling perspective they look similar at the surface. The differences show up in context size, modality support, and which request parameters each one accepts.

Grok 4.20 has the larger context window by a wide margin, at 2,000,000 tokens against 256,000 for Relace Search. Grok 4.20 also accepts image and file inputs alongside text, while Relace Search accepts text only. On output, both return text only. Grok 4.20 supports reasoning and structured outputs, and it exposes parameters such as include_reasoning, reasoning, logprobs, and top_logprobs that Relace Search does not list. Relace Search supports a narrower set: max_tokens, response_format, seed, stop, temperature, tool_choice, tools, and top_p. Both support tool calling, so that is not a differentiator. On price, Relace Search is the cheaper option for input, while Grok 4.20 is cheaper for output.

For practical selection, reach for Relace Search when your workload is plain text, fits inside 256,000 tokens, and you want to keep input costs down, for example bulk classification, extraction, or search-style retrieval prompts where you send a lot of context in and get a short answer back. Reach for Grok 4.20 when you need the long context, when you are feeding in images or files, or when you want reasoning traces or structured outputs. The output price advantage on Grok 4.20 also matters if your responses are long relative to your prompts, since that is where its cheaper output rate pays off.

Calling either model through api14 works the same way, because the gateway exposes an OpenAI-compatible API. You point your client at the api14 base URL, send a standard chat completions request, and set the model field. To use Relace Search, set the model id to "relace/relace-search". To use Grok 4.20, set it to "x-ai/grok-4.20". Everything else in your request body stays the same, so switching between them is a one-line change. Keep in mind that parameters are not fully interchangeable: if you send reasoning or logprobs settings to Relace Search, or rely on image or file content parts, the request will not behave the way it does on Grok 4.20.

There are still cases where Relace Search is the better pick even though Grok 4.20 is larger and more feature rich. If your task is text only, fits within 256,000 tokens, and you are sending far more input than you generate, the cheaper input pricing on Relace Search makes it the more economical choice. It also covers the common parameter set, including tools and response_format, so straightforward tool-calling and JSON-style responses do not require Grok 4.20.

Switch with one string

Both models use the same endpoint and request shape. Change the model value from "relace/relace-search" to "x-ai/grok-4.20" and prompts, streaming and tools keep working.

Relace: Relace Search details & playground →SpaceXAI: Grok 4.20 details & playground →

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api14.com/api/v1",
  apiKey: process.env.API14_API_KEY,
});

const result = await client.chat.completions.create({
  "model": "x-ai/grok-4.20",
  "messages": [
    {
      "role": "user",
      "content": "Explain vector databases in two sentences."
    }
  ]
});
console.log(result.choices[0].message.content);

Relace: Relace Search vs SpaceXAI: Grok 4.20

Specifications

Key differences

In depth

Switch with one string

More comparisons