New: DGC on the Aider polyglot benchmark — 98.9% on a local 27B, same model as every agent See the field →
Benchmark · Aider polyglot

A 27B you run yourself,
solving almost everything

DGC has no time limit — it's built for local models, and it works a hard problem until it's done. Given that room, on a single 27B on your own machine, it solves all but one of the benchmark's 90 problems across six languages.

No time cap — built for local models Matches the field's ceiling A local 27B, not a frontier model
98.9%
89 / 90 solved
qwen3.8:27b-bf16 · pass@2 · run locally
What this is

The harness is the variable, not the model

A coding agent's score is mostly its harness — how it reads, edits, runs tests and recovers — not the model underneath. So we fixed the model to one 27B anyone can run, and measured DGC on the standard Aider polyglot set: 90 exercises, six languages, scored by each one's real test suite, pass@2. DGC's only miss is rust/doubly-linked-list — unsafe-pointer surgery the field-leading harness fails too. The genuine hard tail, not a harness gap.

15/15
Python
15/15
JavaScript
15/15
Java
15/15
Go
15/15
C++
14/15
Rust
The strict-cap view · same model

The safe agent runs a different race

Leaderboards run a tight 600-second-per-problem cap and reward pure sprint speed. DGC optimizes for what the stopwatch can't see: it's built to be pointed at your real repository — permission controls, sandboxing and credential redaction on, running entirely on your own machine, with no clock forcing a slow local model to give up early. Playing on the clock's terms anyway, on the same model, DGC lands level with OpenCode and a couple of problems behind the leaders.

HarnessPass@2 · 600 s capScore
pi
98.9%
Codex CLI
97.8%
DGCthis is us
95.6%
OpenCode
95.6%
goose
94.4%
Aider
77.8%

Lift the cap to the time a local model actually needs — DGC's real design point — and it solves 98.9%. The 600 s sprint is the competitors' home turf; the marathon on your own hardware is ours.

Methodology

How it was measured

Model

Same model, every harness

All agents drive qwen3.8:27b-bf16 through its OpenAI-compatible endpoint. Correctness is hardware-independent.

Tasks

90 problems, 6 languages

15 exercises each in Python, Go, Rust, JavaScript, Java and C++ — a deterministic slice of the Aider polyglot set, pinned to one dataset commit.

Scoring

pass@2, no hints

Round 1 implements and self-tests; if the real suite fails, round 2 gets the failure and retries. Scored only by the exercise's own tests.

Settings

DGC's real settings

No time cap and DGC's shipped iteration budget — exactly what a user gets. The 600 s figures use the leaderboard-standard cap for a like-for-like comparison.

Reproduce it

Run it yourself

The whole harness ships with DGC. Point it at any OpenAI-compatible model and the same numbers fall out — for DGC or any of the other agents.

# get the harness (bundled with DGC)
git clone https://github.com/OpenPeach-ai/dgc && cd dgc/bench
bash install_harnesses.sh        # DGC + the other agents + toolchains

# run DGC on the 90-problem subset
python3 run_bench.py --engine dgc --model qwen3.8:27b-bf16 \
    --base-url http://localhost:11434/v1 --limit 15 --rounds 2

Coming next: SWE-bench — real GitHub issues from Django, SymPy, scikit-learn and more, on the same local model. We'll add the results here.