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.

Nous: Hermes 4 405B vs SpaceXAI: Grok 4.3

A side-by-side comparison of Nous: Hermes 4 405B ($3.50 / 1M tokens) and SpaceXAI: Grok 4.3 ($4.38 / 1M tokens) using current USD prices and capabilities from the api14 catalog. Both run through the same OpenAI-compatible API.

Specifications

	Nous: Hermes 4 405B	SpaceXAI: Grok 4.3
Provider	Nous	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	$0.70 / 1M tokens
Web search	-	$17.50 / 1K searches
Context window	131,072	1,000,000
Max output	117,964	900,000
Input	text	text, image, file
Output	text	text
Tool calling
Structured output
Reasoning control
Added	Aug 26, 2025	Apr 30, 2026

Key differences

Nous: Hermes 4 405B is 20% cheaper on input tokens.
SpaceXAI: Grok 4.3 is 17% cheaper on output tokens.
SpaceXAI: Grok 4.3 reads 7.6× more context (1,000,000 tokens).
Only SpaceXAI: Grok 4.3 supports tool calling.
A request with 1,000 input and 500 output tokens costs $0.00875 with Nous: Hermes 4 405B and $0.00875 with SpaceXAI: Grok 4.3.

In depth

Nous: Hermes 4 405B is made by Nous, while SpaceXAI: Grok 4.3 comes from xAI. Both are text generation models, meaning both take input and produce text output, and both sit in the same broad category on api14. The two differ in release timing, with Hermes 4 405B arriving on 2025-08-26 and Grok 4.3 on 2026-04-30, so Grok 4.3 is the newer of the pair.

The clearest split is context and input handling. Grok 4.3 supports a context window of 1,000,000 tokens, which is larger than the 131,072 tokens Hermes 4 405B accepts. Grok 4.3 also accepts text, image and file inputs, while Hermes 4 405B accepts text only. Both output text only. On tooling, Grok 4.3 is the one that exposes tools and tool_choice, so if your workflow depends on function calling, that is the model to reach for. Hermes 4 405B instead lists a set of sampling and output controls such as frequency_penalty, presence_penalty, repetition_penalty, top_k, temperature, top_p, logprobs, top_logprobs, response_format and structured_outputs. Grok 4.3 supports several of the same controls, including temperature, top_p, logprobs, top_logprobs, response_format, structured_outputs and reasoning, plus reasoning_effort and seed, but it does not list the penalty parameters or top_k. Neither model is described as uniquely better at structured output or reasoning, so treat those as roughly comparable for planning purposes.

For practical selection, start with the input type and the size of the material. If you are sending long documents, large codebases, or anything that needs to fit in a single call, Grok 4.3's larger context makes it the natural choice. If you need to pass images or files alongside text, only Grok 4.3 handles that. If you need tool calling, again Grok 4.3 is the one that supports it. Hermes 4 405B is the better fit when your task is plain text in and text out, when you want the extra sampling controls it exposes (penalties and top_k in particular), or when you are optimizing for input cost, since it is the cheaper of the two on input. Output pricing runs the other way, with Grok 4.3 being cheaper on output, so the cost picture depends on your input to output ratio. For short prompts with long generations, Grok 4.3 may end up cheaper overall; for long prompts with short answers, Hermes 4 405B's cheaper input matters more.

Calling either model through api14 works the same way because the gateway is OpenAI-compatible. You point your client at the same base URL, send the same chat completions style request body, and change only the model id. Use "nousresearch/hermes-4-405b" to route to Hermes 4 405B, or "x-ai/grok-4.3" to route to Grok 4.3. Parameters you pass that a given model does not support should be dropped or ignored, so keep your request payload to the intersection of the two parameter lists if you want one code path to serve both, or branch on the model id when you need tools, reasoning_effort, seed, or the penalty parameters.

There are still cases where Hermes 4 405B is the better pick even though Grok 4.3 covers more ground. If your workload is text only, fits comfortably inside 131,072 tokens, and you care most about input cost, Hermes 4 405B is cheaper on input and gives you penalty and top_k controls that Grok 4.3 does not list. It is also the older, more established option of the two, which may matter if you have already tuned prompts against it.

Switch with one string

Both models use the same endpoint and request shape. Change the model value from "nousresearch/hermes-4-405b" to "x-ai/grok-4.3" and prompts, streaming and tools keep working.

Nous: Hermes 4 405B details & playground →SpaceXAI: Grok 4.3 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-4.3",
  "messages": [
    {
      "role": "user",
      "content": "Explain vector databases in two sentences."
    }
  ]
});
console.log(result.choices[0].message.content);

More comparisons

Hermes 4 405B vs Grok 4.3: API price, context & features · api14