BlogLLM Benchmark Scores as Published, Side by Side
Frontier-Bench: Results as Published
Every result the catalog holds for Frontier-Bench: 4 models from 2 developers, read from 1 page. Each number is as printed by the company named beside it; nothing is ranked or rescored. Models are listed by developer, newest first.
Area Agentic CodingMaintainer Not stated on the page readIntroduced 2026Data version 2026103001, as of 2026-10-30
Frontier-Bench (v0.1 on the Anthropic Claude Opus 5 page) is a set of hard agentic terminal tasks in software and engineering work, such as rebuilding a machine part as a 3D CAD model from a drawing; the model works in a terminal harness and the score is the mean reward over several attempts per task.
Not checked against a maintainer's page.
Frontier-Bench v0.1
| Model | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPT-5.6 SolOpenAI, released 2026-07-09 | 34.4% | Anthropic, Claude Opus 5checked 2026-10-15 | Harness: mini-SWE-agent (Anthropic internal run) |
| Claude Opus 5Anthropic, released 2026-07-24 | 43.3% | Anthropic, Claude Opus 5checked 2026-10-15 | Harness: mini-SWE-agent (Anthropic internal run) |
| Claude Fable 5Anthropic, released 2026-06-09 | 33.7% | Anthropic, Claude Opus 5checked 2026-10-15 | Harness: mini-SWE-agent (Anthropic internal run) |
| Claude Opus 4.8Anthropic, released 2026-05-28 | 21.1% | Anthropic, Claude Opus 5checked 2026-10-15 | Harness: mini-SWE-agent (Anthropic internal run) |
Every number here is read from the same catalog file as the benchmark table, data version 2026103001, and is the number the table shows for that model and test. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.