Grok 4.20 0309 (Reasoning)

Grok 4.20 0309 (Reasoning)

xAI · Released Mar 10, 2026 Professional
Intelligence #10 / 576
82.0 Our Score
Speed #43 / 271
184.1 tokens / sec
Input #482 / 577
$2.00 per 1M tokens
Output #468 / 577
$6.00 per 1M tokens
Context
— Not reported

Analysis Summary

Grok 4.20 0309 (Reasoning) is xAI's reasoning-mode variant of the Grok 4.20 series, released in March 2026. Its intelligence index of 48.5 places it in the excellent tier, supported by a strong GPQA score of 0.885 and a science benchmark of 0.447. The agentic index of 68.7 and terminal benchmark score of 0.409 confirm reliable multi-step tool use and autonomous task handling.

For businesses, this model is well suited to complex reasoning workflows, autonomous coding agents, and research-intensive tasks. The coding index of 42.2 is strong, and the tau2 score of 0.965 suggests high instruction-fidelity in agentic pipelines. Long-context reasoning (lcr 0.59) is above average, making it useful for document-heavy analysis. The main limitation is the absence of vision support in the available data.

At $2 input and $6 output per million tokens, it offers strong price-performance for a model at this capability level. Teams running agentic or reasoning-heavy workloads who want a cost-efficient alternative to the top-tier flagships should consider this model seriously.

Assessed June 6, 2026

Editorial notes

Grok 4.20 0309 in reasoning mode delivers excellent intelligence and strong agentic performance from xAI, with competitive coding capability and very attractive pricing at $2/$6 per million tokens.

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?

Performance Profile

Intelligence7Technical7.1Value6Content8.1
Intelligence 7/10
Technical 7.1/10
Content 8.1/10
Value 6/10

How Grok 4.20 0309 (Reasoning) compares

Grok 4.20 0309 (Reasoning) ranks #26 of 378 AI models we track for overall intelligence, #29 of 315 for coding, #17 of 289 for agentic tasks. At $2.00 per million input tokens it is cheaper than 16% of comparable models.

Performance Indices

Source: Artificial Analysis

48.5 Intelligence Index
42.2 Coding Index
68.7 Agentic Index

Benchmark Scores

Intelligence

GPQA Diamond 88.5% Graduate-level scientific reasoning
HLE 30% Humanity's Last Exam
SciCode 44.7% Scientific computing

Technical

TerminalBench Hard 40.9% Agentic terminal tasks
τ²-Bench 96.5% Conversational agent benchmark

Content

IFBench 82.9% Instruction following
LCR 59% Long-context reasoning

Benchmark data from Artificial Analysis and Hugging Face

How does Grok 4.20 0309 (Reasoning) stack up?

Compare side-by-side with other professional models.

Compare Models

Model Information

ProviderxAI
Release Date March 10, 2026
Status Active

Pricing

Token Type Cost per 1M tokens Cost per 1K tokens
Input $2.00 $0.002000
Output $6.00 $0.006000

Leaderboard Categories

Frequently asked questions about Grok 4.20 0309 (Reasoning)

How much does Grok 4.20 0309 (Reasoning) cost?

Grok 4.20 0309 (Reasoning) costs $2.00 per million input tokens and $6.00 per million output tokens.

Is Grok 4.20 0309 (Reasoning) good for coding?

On our coding benchmark index, Grok 4.20 0309 (Reasoning) ranks #29 of 315 models, placing it in the top quartile of the field for code generation and debugging.

Who created Grok 4.20 0309 (Reasoning)?

Grok 4.20 0309 (Reasoning) is developed by xAI and was released on March 10, 2026.