Skip to main content
BenchLM
Data

ProgramBench binary score (task counts only at 95% or higher) (ProgramBench (binary))

We show this table for reference; we do not rank on it.

Data verified

ProgramBench cleanroom reimplementation tasks scored as pass/fail: a task counts only when its ProgramBench score is at least 95%.

Benchmark score on ProgramBench (binary) — October 29, 2026

We compile the ProgramBench (binary) rows from provider self-reports. Kimi K3.1 Pro leads the table at 31.5%. We do not use these results to rank models overall.

1 modelCodingCurrentDisplay onlyUpdated October 29, 2026

Benchmark score table (1 model)

Benchmark score results for ProgramBench (binary)
RankModel / configurationScoreParameters (B)Open / closed
1Kimi K3.1 ProMoonshot AI
31.5%
Not reportedPending

About ProgramBench (binary)

Year

2026

Tasks

Cleanroom executable reimplementation

Format

Share of tasks at 95% or higher

Difficulty

Whole-program reconstruction

Moonshot reports this binary transform rather than the raw ProgramBench score. BenchLM keeps it on its own key so it never mixes with raw or almost-resolved ProgramBench values.

Freshness and provenance

Version

ProgramBench (binary) 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

Questions

What does ProgramBench (binary) measure?

ProgramBench cleanroom reimplementation tasks scored as pass/fail: a task counts only when its ProgramBench score is at least 95%.

Which model scores highest on ProgramBench (binary)?

Kimi K3.1 Pro by Moonshot AI currently leads with a score of 31.5% on ProgramBench (binary).

How many models are evaluated on ProgramBench (binary)?

1 AI models have published results on ProgramBench (binary) in the BenchLM catalog.

Last updated: October 29, 2026 · BenchLM version ProgramBench (binary) 2026

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 6,100+ subscribers.

One email each week. Unsubscribe anytime.