PRBench Finance leaderboard
- Muse Spark 1.3 leads
- 59.5%
- models measured
- 30
- lead over #2
- +4.5 pts
- median
- 46.4%
Leader on PRBench Finance · read October 22
On October 22, 2026, 30 of our catalog's models had a PRBench Finance result, and Meta Muse Spark 1.3 led with 59.5%, ahead of Meta Muse Spark 1.1 at 55.0%.
PRBench Finance tests complex financial reasoning tasks, including analysis, modeling and decision-making. The number is the score Scale AI publishes for each model.
Role in the index
PRBench Finance is shown as published and does not feed any of our indexes. Read the methodology
| # | Model | Result | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | Muse Spark 1.3Meta | 59.5% | 57.4 | $4.25 | |
| 2 | Muse Spark 1.1Meta | 55.0% | 70.3 | $4.25 | |
| 3 | Claude Fable 5Anthropic | 53.9% | 79.3 | $50.00 | |
| 4 | Claude Opus 4.6Anthropic | 53.3% | 57.2 | $25.00 | |
| 5 | GPT-5OpenAI | 51.3% | 38.8 | $10.00 | |
| 6 | GPT-5 ProOpenAI | 51.1% | Not benchmarked | $120.00 | |
| 7 | Claude Fable 5.1Anthropic | 50.8% | 84.2 | $50.00 | |
| 8 | GPT-5.6 SolOpenAI | 50.4% | 73.5 | $20.00 | |
| 9 | Gemini 4 ArgonGoogle | 49.9% | 81.0 | $10.00 | |
| 10 | Gemini 3.8 FlashGoogle | 49.5% | 61.6 | $3.75 | |
| 11 | o3 ProOpenAI | 49.1% | Not benchmarked | $80.00 | |
| 12 | GPT-5.1OpenAI | 48.0% | 40.4 | $10.00 | |
| 13 | o3OpenAI | 47.7% | Not benchmarked | $8.00 | |
| 14 | GPT-6 AstraOpenAI | 47.5% | 85.1 | $50.00 | |
| 15 | Kimi K2.5Moonshot AI | 46.5% | 38.6 | $2.85 | |
| 16 | GPT-5.2 ProOpenAI | 46.3% | Not benchmarked | $168.00 | |
| 17 | GPT-6.1 SolOpenAI | 46.2% | 82.4 | $10.00 | |
| 18 | Claude Opus 4.5Anthropic | 46.2% | 44.0 | $25.00 | |
| 19 | GPT-5.4OpenAI | 45.6% | 64.4 | $15.00 | |
| 20 | gpt-oss-120bOpenAI | 43.8% | Not benchmarked | $0.50 | |
| 21 | Claude Sonnet 4.5Anthropic | 43.8% | 39.5 | $15.00 | |
| 22 | Kimi K2 ThinkingMoonshot AI | 43.4% | Not benchmarked | $2.50 | |
| 23 | o4 MiniOpenAI | 39.2% | Not benchmarked | $4.40 | |
| 24 | Gemini 2.5 ProGoogle | 38.9% | 19.1 | $10.00 | |
| 25 | Gemini 2.5 FlashGoogle | 38.4% | Not benchmarked | $2.50 | |
| 26 | Kimi K2 0711Moonshot AI | 38.3% | Not benchmarked | $2.30 | |
| 27 | Claude Opus 4.1Anthropic | 35.2% | Not benchmarked | $75.00 | |
| 28 | GPT-4.1OpenAI | 34.3% | Not benchmarked | $8.00 | |
| 29 | R1 0528DeepSeek | 32.7% | Not benchmarked | $2.18 | |
| 30 | GPT-4.1 MiniOpenAI | 30.4% | Not benchmarked | $1.60 |
Only models in our catalog are listed, and each appears once with its best published run. 66 rows were read from the source on October 22, 2026, and rows that match no catalog model are left out.