Omen Alpha Benchmarks
Independent coding benchmark results for omen-alpha — scores, cost per prompt, and speed, with the visible methodology and third-party context behind every number.
The benchmark snapshot.
One dated leaderboard result, shown with its component project scores rather than a made-up composite benchmark.
Published result · September 4, 2026 · OpenCode · Methodology v1
Coding benchmark scores.
The recorded OpenCode evaluation scores four implementation projects on a five-point project rubric. The leaderboard reports a 23.14 total across the four visible project categories, plus a separate code-quality component in the expanded view.
| Benchmark project | Omen Alpha | Max | What the score represents |
|---|---|---|---|
| Code quality | 9.94 | 20 | Expanded leaderboard code-quality component |
| CSV import (PHP) | 4 | 5 | Project-specific implementation score |
| Offline sync (PHP) | 3.5 | 5 | Project-specific implementation score |
| Bank feed (Dart / Flutter) | 2.7 | 5 | Project-specific implementation score |
| Shipping quotes (Go) | 3 | 5 | Project-specific implementation score |
The project rows and 23.14 leaderboard total are transcribed from the September 4, 2026 OpenCode snapshot. The code-quality value uses the expanded leaderboard view; the displayed component scales are not summed together as a new score.
Omen Alpha vs Ox Alpha.
A fair comparison needs the same prompts, model settings, evaluator, and date. No directly comparable Ox Alpha run is published in the source snapshot, so the table below marks the missing evidence instead of inventing a result.
| Metric | Omen Alpha (High) | Ox Alpha |
|---|---|---|
| Coding score | 23.14 / 40 | No same-run result published |
| Avg cost / prompt | $0.03 | Not disclosed in this dataset |
| Speed / prompt | 01:51 | Not disclosed in this dataset |
| Evaluation environment | OpenCode | Not supplied |
Full Omen Alpha vs Ox Alpha comparison → will be updated when a reproducible, source-backed Ox Alpha run is available.
Benchmark cost & speed.
The snapshot reports an average cost of $0.03 per prompt and an average elapsed time of 01:51 per prompt. These are observed benchmark-run metrics, not a promise of latency or cost for every workload.
Listed token pricing: input $0.20 · output $0.66 · cached read $0.04 per million tokens. Confirm current commercial terms with Tokenra before deployment. A cache ratio was not included in the supplied leaderboard snapshot, so it is reported as unavailable rather than estimated.
Methodology.
- Evaluation setFour coding projects shown in the OpenCode leaderboard: CSV import (PHP), offline sync (PHP), bank feed (Dart / Flutter), and shipping quotes (Go). Snapshot date: September 4, 2026.
- SetupTested with OpenCode using the Omen Alpha (High) configuration. Temperature, max tokens, hardware, and concurrency were not disclosed in the supplied source.
- ScoringProject results use a five-point rubric per project. The expanded leaderboard also reports code quality on a 20-point scale. The source does not publish the individual test assertions or grader weights.
- ReproducibilityThis page preserves the visible source values and labels unknowns explicitly. Prompts, raw logs, and a downloadable JSON/CSV result file are not currently published here.
Third-party results.
Independent and community sources can add useful context, but their dates, prompts, and evaluation conditions may differ from the OpenCode snapshot above.
- OpenCode Data — omen-alpha usage & rankLeaderboard source referenced for the displayed run.
- AI Coding Daily — Omen Alpha coding benchmarkIndependent model coverage and benchmark context.
Frequently asked questions.
How does Omen Alpha compare to Ox Alpha on benchmarks?
This page records Omen Alpha's OpenCode leaderboard result. A directly comparable Ox Alpha run with the same dataset, rubric, and environment has not been published, so a like-for-like score comparison is not claimed.
What does Omen Alpha cost per prompt?
The benchmark snapshot reports an average cost of $0.03 per prompt. Current listed token pricing is $0.20 input, $0.66 output, and $0.04 cached read per million tokens; confirm live pricing with Tokenra before deployment.
Is there an official Omen Alpha benchmark?
The result shown here is a dated OpenCode leaderboard snapshot, not a vendor-published benchmark suite. The evaluation date, environment, and visible scoring rubric are documented on this page.
Who built Omen Alpha?
Omen Alpha is an unannounced coding model; its vendor has not been publicly confirmed. Community testing suggests a possible GLM lineage, but that is not an official endorsement or attribution.
Run the model yourself.
Get an API key through Tokenra and test your own representative coding workload.