cd /news/artificial-intelligence/solving-hack-the-box-challenges-with… · home topics artificial-intelligence article
[ARTICLE · art-94337] src=theaq.blog ↗ pub= topic=artificial-intelligence verified=true sentiment=↑ positive

Solving Hack the Box Challenges with Grok 4.6

XAI's Grok 4.6 achieved an 80% benchmark score on the HTB-Challenger Benchmark, solving 14 of 16 Hack The Box challenges with zero false positives, surpassing all previously tested models in both speed and cost. The model, released today, completed challenges in a median of 14 steps and $0.23 per challenge, with a total cost of $17.31 across 16 challenges. The author, who previously called an older version 'useless', now regards Grok 4.6 as the best model tested on offensive-security challenges.

read4 min views1 publishedAug 12, 2026
Solving Hack the Box Challenges with Grok 4.6
Image: source

When I tested an older version of Grok this spring, my only comment was blunt: “grok-4.20 was useless.” So when Grok 4.5 was released a month ago and started receiving very positive reviews, I was quite surprised.

Today, Grok 4.6 was released and I’ve finally had a chance to test it against HTB challenges, and the team behind it deserves an apology. I’m not sure what the folks at xAI did, but the performance increase over the older version is jaw-dropping.

This blog post is part of a series of tests for the HTB-Challenger Benchmark. See the

[benchmark results page]for all results and the[benchmark methodology]to learn how the benchmark is calculated.

During my testing, I watched this model crack one HTB challenge after another in a steady, confident way, and the final boss fight against the last two challenges was exhilarating, even though it was ultimately unsuccessful. In the end, with a 80% score, it’s the undisputed new king of my leaderboard - not only because of its results, but also because of its speed (it had the fewest steps per challenge of all the models I’ve tested so far) and cost. Look at the Cost vs. Benchmark Score graph on the HTB-Challenger Benchmark page to see what I mean.

So let’s forget that useless old version. For me, Grok 4.6 is currently the best model I’ve tested on offensive-security challenges.

Overall benchmark results #

Number of challenges: 16Number of solved challenges: 14Number of false positives: 0Runs where the model gave up: 0Runs that reached the step or cost limit: 2Runs where the model got stuck: 0Benchmark score: 80.0%

Metric Per challenge (median) Total
Model steps 14 452
Model cost $0.23 $17.31
Duration 00:03:38 02:45:35
Number of input tokens 0.23M 20.29M
Number of output tokens 0.01M 0.47M
Number of read_file tool calls
1.0 41
Number of write_file tool calls
0.0 35
Number of execute_command tool calls
10.0 403
Number of web_search tool calls
0.0 15

Results by challenge difficulty #

Very Easy challenges

Number of challenges: 4Number of solved challenges: 4Number of false positives: 0Runs where the model gave up: 0Runs that reached the step or cost limit: 0Runs where the model got stuck: 0Benchmark score: 100.0%

Metric Per challenge (median) Total
Model steps 11.5 50
Model cost $0.10 $0.96
Duration 00:01:26 00:09:26
Number of input tokens 0.10M 1.05M
Number of output tokens 0.00M 0.03M
Number of read_file tool calls
0.5 8
Number of write_file tool calls
0.0 1
Number of execute_command tool calls
9.0 42
Number of web_search tool calls
0.5 4

Easy challenges

Number of challenges: 4Number of solved challenges: 4Number of false positives: 0Runs where the model gave up: 0Runs that reached the step or cost limit: 0Runs where the model got stuck: 0Benchmark score: 100.0%

Metric Per challenge (median) Total
Model steps 11.5 43
Model cost $0.14 $0.58
Duration 00:03:00 00:12:29
Number of input tokens 0.12M 0.51M
Number of output tokens 0.01M 0.03M
Number of read_file tool calls
1.0 4
Number of write_file tool calls
2.0 8
Number of execute_command tool calls
8.0 30
Number of web_search tool calls
0.5 3

Medium challenges

Number of challenges: 4Number of solved challenges: 4Number of false positives: 0Runs where the model gave up: 0Runs that reached the step or cost limit: 0Runs where the model got stuck: 0Benchmark score: 100.0%

Metric Per challenge (median) Total
Model steps 19 106
Model cost $0.58 $3.98
Duration 00:09:39 00:45:15
Number of input tokens 0.61M 4.50M
Number of output tokens 0.02M 0.11M
Number of read_file tool calls
0.0 4
Number of write_file tool calls
0.5 6
Number of execute_command tool calls
16.5 102
Number of web_search tool calls
0.0 0

Hard challenges

Number of challenges: 4Number of solved challenges: 2Number of false positives: 0Runs where the model gave up: 0Runs that reached the step or cost limit: 2Runs where the model got stuck: 0Benchmark score: 50.0%

Metric Per challenge (median) Total
Model steps 63.5 253
Model cost $2.89 $11.79
Duration 00:24:56 01:38:25
Number of input tokens 3.50M 14.23M
Number of output tokens 0.08M 0.31M
Number of read_file tool calls
4.5 25
Number of write_file tool calls
1.0 20
Number of execute_command tool calls
55.0 229
Number of web_search tool calls
1.0 8
── more in #artificial-intelligence 4 stories · sorted by recency
── more on @xai 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/solving-hack-the-box…] indexed:0 read:4min 2026-08-12 ·