NOUMENOpen app
NOUMEN

xAI · Balanced

Grok 4.3

Added Apr 30, 2026

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Use in the app

Grok 4.3 is SpaceXAI's reasoning model for text and image inputs, built for agentic workflows, instruction following, factual accuracy, long-document analysis, and deep research. Reasoning is always active and requests above 200k total tokens are charged at the higher long-context rate. Content policy rejections can still be charged: SpaceXAI may pass through a $0.05 moderation-failure fee, or a $0.055 usage-guidelines violation fee, depending on which rejection upstream returns.

  • Reasoning
  • Tools
  • Vision
  • Structured output
Input
Text, Images
Output
Text
Discovery tier
Balanced

Pricing

USD per 1M tokens
Input
$1.25
Output
$2.50
Cache read
$0.20

Reference prices as published by the model provider. API example

Specifications

Context window
1M
Max output
—
Average output (7 days)
466 tokens
Longer than 46% of 473 models

Side by side

Grok 4.3 compared

Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Grok 4.3xAIThis modelQwen3.5 9BQwenRemove
ProviderxAIQwen
Discovery tierBalancedOpen weights
AddedApr 30, 2026Mar 10, 2026
Limits
Context window1MBest262.14K
Max output—65.54K
Average output466 tokens1.4K tokens
Pricing, USD per 1M tokens
Input$1.25$0.05Best
Output$2.50$0.15Best
Cache read$0.20—
Performance
Throughput357.8 tok/s—
Time to first token17.5 s—
Request success (24h)100.0%Best99.8%
Benchmark indexes
Intelligence Index14.0Best13.3
Coding Index35.2Best23.5
Capabilities
ReasoningYesYes
ToolsYesYes
VisionYesYes
Structured outputYesNo
Video inputNoYes
Reasoning effort——

Public evaluations

Benchmarks

Independent evaluation results, with how each score ranks among the models in the Noumen catalog.

  • Grok 4.3
  • Qwen3.5 9B

Intelligence Index

  • Grok 4.314.0
  • Qwen3.5 9B13.3

Grok 4.3: Better than 35% of 326 scored models

3.8Catalog models by score55.1

Indexes

  • Coding Index

    Composite of the coding evaluations

    • Grok 4.3: 35.2Top
    • Qwen3.5 9B: 23.5

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Grok 4.3: 0.9%
    • Qwen3.5 9B: —

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Grok 4.3: 0.2%
    • Qwen3.5 9B: —

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Grok 4.3: 724 Elo
    • Qwen3.5 9B: —

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Grok 4.3: 941 Elo
    • Qwen3.5 9B: —

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Grok 4.3: 2.8%
    • Qwen3.5 9B: —

  • AA-LCRv1.1

    Reasoning across long documents

    • Grok 4.3: 32.3%
    • Qwen3.5 9B: 46.0%Top

  • MLCR-AA

    Long-context reasoning over medical records

    • Grok 4.3: 5.6%
    • Qwen3.5 9B: —

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Grok 4.3: 6.8%
    • Qwen3.5 9B: 9.4%Top

  • GPQA Diamond

    Graduate-level science questions

    • Grok 4.3: 65.8%
    • Qwen3.5 9B: 78.6%Top

  • IFBench

    Following precise output instructions

    • Grok 4.3: 47.6%Top
    • Qwen3.5 9B: 37.8%

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Grok 4.3: 0.0%
    • Qwen3.5 9B: —

  • SciCode

    Scientific computing in Python

    • Grok 4.3: 39.4%
    • Qwen3.5 9B: —

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • Grok 4.3: 18.9%Top
    • Qwen3.5 9B: 18.2%

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Grok 4.3: 65.8%
    • Qwen3.5 9B: 85.1%Top

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Grok 4.3: 32.3%
    • Qwen3.5 9B: 46.0%Top

Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis

Live routing data

Performance and reliability

Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.

Throughput
357.8 tok/s
Faster than 95% of 260 models
Time to first token
17.5 s
Quicker than 4% of 260 models
Prompt cache hits
19.9%
About $1.04 per 1M input tokens with caching
Request success (24h)
100.0%
Reliable

Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.

Hourly request success · last 72 hoursLowest hour 100.0%
  • Healthy (98% or more)
  • Degraded (90–98%)
  • Down (under 90%)
  • No requests

Developers

Use it through the API

Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.

Read the quickstart
curl https://www.noumen.xyz/api/v1/chat/completions \
  -H "Authorization: Bearer $NOUMEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "x-ai/grok-4.3",
    "messages": [{ "role": "user", "content": "Explain this stack trace." }]
  }'
import os
from openai import OpenAI

client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
    model="x-ai/grok-4.3",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

xAI

Other xAI models, starting with the same family.