Samaya AI, 2026
FrontierFinance
Open-ended investment research questions across six use cases: screening, company research, sector and macro, data extraction, earnings and events, and catalyst monitoring. Answers are scored against expert-written, source-attributed rubrics.
About this benchmark
- Size
- 220 queries, 11,543 rubric items
- Public split
- All 220 queries and rubrics, on Hugging Face (gated with automatic approval; you agree not to reshare)
- Grading
- An LLM judge panel decides each rubric item (majority vote, first judge breaks ties). Recommended judges are Claude Sonnet 4.6, Gemini 3.1 Pro and GPT 5.4. Headline metric is the rubrics qualification rate.
- Licence
- CC-BY-4.0 (data); Apache-2.0 (grader code)
- Submission
- Not stated
- Products allowed
- Yes
- Latest version
- 2026-10-28
- What it does not cover
- No trade execution, portfolio outcomes or advice-safety checks.
Results
10 modelsEach score shows who reported it: the benchmark's authors (quoted from their paper), the benchmark's official board, or Moneda (run with the benchmark's own grader and rules). We do not copy other aggregators.
Best value: Samaya Sterling, 58.9 at $0.610 per question (cheapest within 5% of the top score)
FrontierFinance: score vs cost
Score against cost per test. Up and to the left is better.
Updated 28 Oct 2026
FrontierFinance 2026-10-28
rubrics qualification rate %, higher is better. 10 models.
Updated 28 Oct 2026
- Samaya Sterling58.9
- Legacy Samaya System (high effort)56
- Claude Fable 5.1 · Finance Agent v255.9
- Muse Spark Open · Finance Agent v253.7
- Legacy Samaya System52.9
- Muse Wonder 1.0 · Finance Agent v252.6
- Claude Opus 5.5 · Finance Agent v250.9
- Gemini 3.8 Flash · Finance Agent v250
- Claude Fable 5 · Finance Agent v249.2
- GPT 5.6 Sol · Finance Agent v246.8
FrontierFinance: cost
cost per test item (USD), lower is better. 10 models.
Updated 28 Oct 2026
- Samaya Sterling0.610
- Legacy Samaya System0.930
- Gemini 3.8 Flash · Finance Agent v21.2
- Muse Spark Open · Finance Agent v21.4
- Legacy Samaya System (high effort)1.8
- Claude Opus 5.5 · Finance Agent v21.9
- GPT 5.6 Sol · Finance Agent v23.0
- Muse Wonder 1.0 · Finance Agent v23.4
- Claude Fable 5 · Finance Agent v24.1
- Claude Fable 5.1 · Finance Agent v26.9
FrontierFinance 2026-10-28
rubrics qualification rate %, higher is better. 10 models.
Updated 28 Oct 2026
- 1Samaya Sterling58.9
- 2Legacy Samaya System (high effort)56
- 3Claude Fable 5.1 · Finance Agent v255.9
- 4Muse Spark Open · Finance Agent v253.7
- 5Legacy Samaya System52.9
- 6Muse Wonder 1.0 · Finance Agent v252.6
- 7Claude Opus 5.5 · Finance Agent v250.9
- 8Gemini 3.8 Flash · Finance Agent v250
- 9Claude Fable 5 · Finance Agent v249.2
- 10GPT 5.6 Sol · Finance Agent v246.8
| 1 | Samaya Sterling | Samaya AI | 58.9 | Samaya harness | Samaya leaderboard Official board, reported 28 Oct 2026 · checked 1 Nov 2026 |
|---|---|---|---|---|---|
| 2 | Legacy Samaya System (high effort) | Samaya AI | 56 | Samaya harness | Samaya leaderboard Official board, reported 28 Oct 2026 · checked 1 Nov 2026 |
| 3 | Claude Fable 5.1 · Finance Agent v2 | Anthropic | 55.9 | Finance Agent v2 harness | Samaya leaderboard Official board, reported 28 Oct 2026 · checked 1 Nov 2026 |
| 4 | Muse Spark Open · Finance Agent v2 | Meta | 53.7 | Finance Agent v2 harness | Samaya leaderboard Official board, reported 28 Oct 2026 · checked 1 Nov 2026 |
| 5 | Legacy Samaya System | Samaya AI | 52.9 | Samaya harness | Samaya leaderboard Official board, reported 28 Oct 2026 · checked 1 Nov 2026 |
| 6 | Muse Wonder 1.0 · Finance Agent v2 | Meta | 52.6 | Finance Agent v2 harness | Samaya leaderboard Official board, reported 28 Oct 2026 · checked 1 Nov 2026 |
| 7 | Claude Opus 5.5 · Finance Agent v2 | Anthropic | 50.9 | Finance Agent v2 harness | Samaya leaderboard Official board, reported 28 Oct 2026 · checked 1 Nov 2026 |
| 8 | Gemini 3.8 Flash · Finance Agent v2 | 50 | Finance Agent v2 harness | Samaya leaderboard Official board, reported 28 Oct 2026 · checked 1 Nov 2026 | |
| 9 | Claude Fable 5 · Finance Agent v2 | Anthropic | 49.2 | Finance Agent v2 harness | Samaya leaderboard Official board, reported 28 Oct 2026 · checked 1 Nov 2026 |
| 10 | GPT 5.6 Sol · Finance Agent v2 | OpenAI | 46.8 | Finance Agent v2 harness | Samaya leaderboard Official board, reported 28 Oct 2026 · checked 1 Nov 2026 |
Papers and links
Last checked 1 Nov 2026. Something out of date or wrong? Send a correction.
