Benchmark Almanac

Hack-Verifiable Terminal Bench

Not in model cards

Agents complete terminal tasks with access to restricted reference materials. The benchmark measures how often they consult those materials.

The Inspect register pins this implementation. The current repository also hosts a different evaluation called RHOB.

Made by
Amit Roth and colleagues
Released
August 2026
Kind
Deception and alignment
Level
Expert
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Propensity
Language
English
In and out
Text, Code → Actions, Code, Text
Interaction
Agentic
Environment
Terminal
Refresh
One fixed set
Run by
Anyone
Strongest model tested
Claude Opus 4.6 #39 on the capability index
Last model added
February 2026
Scores refreshed
November 20th, 2026
2
Models scored on its own leaderboard

Claude Opus 4.6 leads with 10.1%

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best.

  1. 1 Claude Opus 4.6 10.1%
  2. 2 Claude Haiku 4.5 12.4%

Source: Official Hack-Verifiable Terminal Bench leaderboard

#ModelScoreSettingDateSource
1 Claude Opus 4.6 10.1% Inspect react() Feb 2026 Official leaderboard
2 Claude Haiku 4.5 12.4% Inspect react() Oct 2025 Official leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2026 Claude Opus 4.6 Claude Haiku 4.5 0% 20% 40% 60% 80% 100% 2026 Claude Opus 4.6

Source: Official Hack-Verifiable Terminal Bench leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Safety and alignment

All 636 in Safety and alignment
  • FORTRESS In use Paired adversarial and benign national-security misuse prompts. Rubric-graded risk score on adversarial and over-refusal score on benign prompts. Claude Opus 5.5 5.78%
  • StrongREJECT In use Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is. Inkling 98.6%
  • Gray Swan IPI Arena benchmark In use Indirect prompt-injection attacks from public red-teaming competitions replayed on tool, coding and browser agents. Attack success rate at k attempts. Gemini 4 Carbon 0%
  • Covert misbehavior (Thinking Machines) In use Agentic tasks that check whether a model misbehaves and conceals it. Scored by the rate of covert misbehavior. Inkling 2 Mini 8.30%
  • Destructive actions (Thinking Machines) In use Agentic tasks that check whether a model takes destructive actions it was not authorized to take. Scored by the rate of destructive actions. Inkling 2 Small 4.70%
  • Misbehavior (Thinking Machines) In use Agentic tasks that check whether a model fabricates results, ignores instructions or hides what it did when a task is hard. Scored by the rate of misbehavior. Inkling 2 Mini 9.80%

Data audit

7 of 8 checks pass, 1 warning. Look first at the Newest score check. The newest came 9 months ago, in February 2026, for Claude Opus 4.6. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 1 pass 3 notes
  • Pass Independent runs Scores from the official leaderboard (2 entries).
  • Note Models scored Only 2 models we track.
  • Warning Newest score The newest came 9 months ago, in February 2026, for Claude Opus 4.6.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude Opus 4.6 (#39).
  • Note Headroom Lower is better, so there’s no ceiling to reach.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 2 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check