Inference
EN
xAI

Grok 4.20

Grok 4.20 is a reasoning model from xAI. 18 benchmarks count toward its index, in 6 categories.

  • Closed weights
  • Reasoning
  • Input: text · image · file
  • Released 2026-03
RankedEnough results for an index.
IndexAll categories, out of 100.52.8±4.8Median 51.0
CoverageIndex weight that has results.90%
SpeedOutput tokens per second.72/sMedian 67/s
First tokenSeconds to the first output token.0.7 sMedian 1.3 s
Input / 1MUS dollars per 1M input tokens.$1.25Median $0.60
Output / 1MUS dollars per 1M output tokens.$2.50Median $2.55
ContextMaximum tokens in one request.2MMedian 262K
EloLMArena rating and rank.—Median 1422

Details

From OpenRouter
API IDThe model ID on OpenRouter.
x-ai/grok-4.20
Max outputMaximum output tokens per request.
1.8M
Knowledge cutoffLast date of training data.
1 Sep 2025
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
$0.20
WeightsLink to the published weights.
—
RetiresDate OpenRouter removes the model.
—

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.4 counted50.7122 / 218
CodingCode writing and repair.2 counted55.492 / 209
ReasoningLogic problems and puzzles.1 counted56.0—
MultimodalTasks with images and text.4 counted45.872 / 84
KnowledgeFacts and expert knowledge.4 counted55.494 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.None counted——
MathMath problems.3 counted57.455 / 168

Results

18 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
GPQA diamond · Epoch AIKnowledge89.3% ±1.948 / 217
OTIS Mock AIME 2024-2025Math92.2% ±3.354 / 200
Software Engineering Benchmark VerifiedCoding76.7%40 / 114
MazeBench: visual spatial reasoning in a 3D open worldAgentic0.0%23 / 48
GPQA Diamond · BenchLMKnowledge88.5%43 / 84
Chess PuzzlesReasoning29.0% ±4.651 / 90
FrontierMath-Tiers-1-3-v2-PrivateMath44.9% ±3.059 / 90
Massive Multi-discipline Multimodal Understanding ProMultimodal75.2%36 / 53
Humanity's Last Exam without toolsKnowledge31.6%36 / 53
FrontierMath-Tier-4-v2-PrivateMath17.1% ±5.952 / 72
SimpleQA VerifiedKnowledge30.2% ±1.569 / 91
SimpleVQAMultimodal57.4%14 / 18
Terminal-Bench 2.0 · BenchLMAgentic47.1%32 / 41
SWE-bench ProCoding51.8%65 / 83
Gert Labs Composite Game BenchmarkAgentic38.4%42 / 52
ERQAMultimodal54.1%15 / 18
HealthBench HardKnowledge20.3%12 / 13
DeepSearchQAAgentic62.8%23 / 24
CharXiv ReasoningMultimodal60.9%48 / 49
LiveCodeBench ProCoding74.2%—
MedXpertQA MultimodalMultimodal65.8%—
MedXpertQA TextKnowledge50.2%—

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseEpoch AICC BY

History

5 releases
Index53.5 → 52.8
23 Oct 202612 Nov 2026
Output / 1M$2.50 → $2.50
23 Oct 202612 Nov 2026

Questions

How much does Grok 4.20 cost?
$1.25 per 1M input tokens, $2.50 per 1M output tokens.
How good is Grok 4.20?
Its index is 52.8 out of 100. That is place 121 of 283 ranked models.
What is Grok 4.20 best at?
Math: 57.4 out of 100, place 55 of 168.
How fast is Grok 4.20?
72 output tokens per second, 0.7 s to the first token.
What is the context window of Grok 4.20?
2M tokens.
Is Grok 4.20 open source?
No. Its weights are not published.
When was Grok 4.20 released?
March 2026.

Badge

Grok 4.20 on Inference360: place 121 of 283, index 52.8
HTML
<a href="https://inference360.ai/models/grok-4-20-beta"><img src="https://inference360.ai/badges/grok-4-20-beta.svg" alt="Grok 4.20 on Inference360: place 121 of 283, index 52.8" height="20"></a>
Markdown
[![Grok 4.20 on Inference360: place 121 of 283, index 52.8](https://inference360.ai/badges/grok-4-20-beta.svg)](https://inference360.ai/models/grok-4-20-beta)

Compare with

Kimi K2.5 (Reasoning)53.0dots3-note Preview63.4

Lower price

dots3-note Preview63.4 · FreeApodex 1.1 Mini55.7 · FreeNemotron 3 Ultra50.8 · Free

More from xAI

All models
Grok 4.20 Multi-agentUnrankedGrok 4.353.3Grok Build 0.152.7Grok RealtimeUnrankedGrok 4.1 Fast (Reasoning)47.3