FrontierCode 2.0, Artificial Analysis evaluation (FrontierCode 2.0 (AA))
We show this table for reference; we do not rank on it.
Artificial Analysis's own run of Cognition's FrontierCode 2.0 coding-agent benchmark, one of the nine evaluations in its Intelligence Index v5, judged on whether each pull request is mergeable.
Benchmark score on FrontierCode 2.0 (AA) — November 11, 2026
We compile the FrontierCode 2.0 (AA) rows from provider self-reports. Claude Haiku 5.5 leads the table at 34.9%, followed by Muse Ember (27.2%) and GPT-6 Luna (25.2%). We do not use these results to rank models overall.
Claude Haiku 5.5
Anthropic
Muse Ember
Meta
GPT-6 Luna
OpenAI
3 modelsCodingCurrentDisplay onlyUpdated November 11, 2026
Benchmark score table (3 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Claude Haiku 5.5Anthropic | 34.9% | Not reported | Closed |
| 2 | Muse EmberMeta | 27.2% | Not reported | Closed |
| 3 | GPT-6 LunaOpenAI | 25.2% | Not reported | Closed |
Among the reported FrontierCode 2.0 (AA) rows, Claude Haiku 5.5 is first at 34.9%. The third row is 9.7 points behind. The broader top-10 range is 9.7 points, so many of the published results sit in a relatively narrow band.
3 models have been evaluated on FrontierCode 2.0 (AA). The benchmark falls in the Coding category. FrontierCode 2.0 (AA) is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About FrontierCode 2.0 (AA)
Year
2026
Tasks
132 private software-engineering tasks
Format
Weighted rubric score with blocking criteria
Difficulty
Frontier coding-agent quality
Artificial Analysis runs FrontierCode 2.0 under its own harness, so its scores differ from the Cognition-published FrontierCode 2.0 table that BenchLM stores on the FrontierCode 2.0 page (for example Claude Haiku 5.5 at 34.9 here and 27.7 there). BenchLM first added this lane from Meta's Muse Ember launch post, which quotes the Artificial Analysis values, and keeps it display-only.
Freshness and provenance
Version
FrontierCode 2.0 (AA) 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
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 FrontierCode 2.0 (AA) measure?
Artificial Analysis's own run of Cognition's FrontierCode 2.0 coding-agent benchmark, one of the nine evaluations in its Intelligence Index v5, judged on whether each pull request is mergeable.
Which model scores highest on FrontierCode 2.0 (AA)?
Claude Haiku 5.5 by Anthropic currently leads with a score of 34.9% on FrontierCode 2.0 (AA).
How many models are evaluated on FrontierCode 2.0 (AA)?
3 AI models have published results on FrontierCode 2.0 (AA) in the BenchLM catalog.
Compare top models on FrontierCode 2.0 (AA)
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.