Benchmarks / Professional Domains
CorpFin
Vals AI's corporate-finance document analysis benchmark.
maintained by Vals AI · www.vals.ai · 122 observations on record
Versions
1 version
| Version | Tasks |
|---|---|
| CorpFin current corp-fin | — |
Current leaderboard
CorpFin · Accuracy · top 25 of 122 systems
| # | System | Accuracy % |
|---|---|---|
| 1 | Claude Fable 5max effort Anthropic | 71.8% |
| 2 | Muse Spark 1.1xhigh effort Meta AI | 71.3% |
| 3 | Grok 4.3 xAI | 68.5% |
| 4 | GPT 5.5 (2026-04-22)xhigh effort OpenAI | 68.4% |
| 5 | Kimi K2 5thinking Moonshot AI | 68.3% |
| 6 | Minimax M3 MiniMax | 68.1% |
| 7 | Qwen3 Max (2026-01-23) Alibaba | 68.0% |
| 8 | Grok 4.5high effort xAI | 67.4% |
| 9 | Claude Opus 4.6max effort Anthropic | 67.0% |
| 10 | Claude Sonnet 5max effort Anthropic | 67.0% |
| 11 | Grok 4 Fast Reasoning xAI | 66.9% |
| 12 | Kimi K2 6 Moonshot AI | 66.7% |
| 13 | Claude Opus 4.8max effort Anthropic | 66.7% |
| 14 | Qwen3 6 Max Preview Alibaba | 66.5% |
| 15 | Gemini 3 Flash Previewhigh effort Google DeepMind | 66.4% |
| 16 | GLM 5.2max effort Z.ai | 66.1% |
| 17 | Claude Opus 4.7max effort Anthropic | 66.1% |
| 18 | Grok 4 0709 xAI | 66.0% |
| 19 | Grok 4.1 Fast Reasoning xAI | 66.0% |
| 20 | GPT 5.2 (2025-12-11)xhigh effort OpenAI | 65.9% |
| 21 | Nemotron 3 Ultra 550B A55b NVIDIA | 65.5% |
| 22 | Claude Sonnet 4.6max effort Anthropic | 65.3% |
| 23 | Qwen3 5 Plusthinking Alibaba | 65.3% |
| 24 | GPT 5.6 Terraxhigh effort OpenAI | 65.3% |
| 25 | GPT 5.4 (2026-03-05)xhigh effort OpenAI | 65.3% |
One row per evaluated system — reasoning-effort variants rank separately.