Benchmark definition
Terminal-Bench 3.0
3.0Applicability does not mean evaluated; benchmark result/model score data is outside this UI contract.
About
- Description
- Terminal-Bench 3.0
- What it measures
- Terminal-Bench 3.0
- Comparison guidance
- Not published
- Intended difficulty
- Not published
Not recorded for this benchmark: what it does not measure, why it matters, motivation, gap addressed.
Classification
- Categories
- Terminal AgentAgentic
- Domains
None recorded
- Evaluation style
None recorded
Structure
One undivided item set. No components or scenarios recorded.
Tasks
Metrics
What a result on this benchmark would be reported as. ctxdex does not carry results.
| Metric | Role | Direction | Unit |
|---|---|---|---|
Accuracy | Primary | Higher is better | Percent |
Pass@k Probability/estimate that at least one of k generated solutions passes the benchmark correctness tests. | Secondary | Higher is better | Percent |
Methodology
Continuous semantic-versioned benchmark family. Material task/resource/evaluator changes cross release comparability boundaries and may require reruns.
- Source lane
- L4
Evaluation environment
What the benchmark's harness provides and allows. These are properties of the environment, not model requirements, and applicability does not test for them.
- Actions
- Terminal commandPTY keystroke
- Observations
- Terminal text
- Session
- Both exec and PTY
- Reset
- Recreate
- Network
- External allowed
- State persists per
- Episode
Shell/PTY wrapper is harness-specific unless a benchmark scenario says otherwise.
- Evaluation style
- Not recorded
- Environment summary
- Not published
Model requirements
Required conditions decide applicability. Relevant ones describe what the benchmark exercises and decide nothing.
Required · decides applicability
Relevant · informational
CTXDEX models
51 of the 58 active, preview, experimental models meet this benchmark’s recorded requirements, at some scope. Applicable means the requirements are met, not that the model was evaluated.
View all 51 applicable models →- Applicable · full
- 51
- Applicable · components
- 0
- Applicable · scenarios
- 0
- Conditionally applicable
- 0
- Not recorded
- 7
Counted over 58 models in the default lifecycle scope (active, preview, experimental).
Every required condition, for the whole benchmark, is recorded as supported. Each line is one recorded condition and why it counts.
Every one of the 1 required conditions is recorded as supported.
MODALITY:INPUT:TEXTthe catalogue records SUPPORTED for TEXT (input)
Every one of the 1 required conditions is recorded as supported.
MODALITY:INPUT:TEXTthe catalogue records SUPPORTED for TEXT (input)
Applicability/relevance ordering only; never a performance ranking.
Scale
- 74TaskTotal · Exact
Data and access
Coverage
- Languages, regions
None recorded
- Coverage summary
- Not published
Limitations
None recorded
Lineage
- Family
- FamilyTerminal-Bench
- Related
- Other definitions in family
- Terminal-Bench 1.0 1.0 SupersededTerminal-Bench 2.0 2.0 SupersededTerminal-Bench 2.1 2.1 SupersededTerminal-Bench 4.0 4.0 Active
- Changes
None recorded
Sources
- https://www.tbench.ai/benchmarks ↗Official website · Benchmark official
- https://www.tbench.ai/news/terminal-bench-3-0 ↗Official website · Benchmark official
Freshness
- Content version
- B2-1.0
- Last reviewed
- 2026-09-23
- Refresh model
- Unknown
- Time sensitivity
- Unknown
- Benchmark data
- c40cccac878af5d0
13 contract fields are empty for every benchmark in this release and are not shown: applicability groups, conditional requirements, feature requirements, model relevance domains, required capabilities, openness summary, reference artifacts, interaction protocols, tool protocols, benchmark last major refresh at, short name, target expertise level, target population.