Scale Labs PRBench Finance (PRBench Finance)
We show this table for reference; we do not rank on it.
A Scale Labs public leaderboard mirrored as display-only reference data. It does not affect BenchLM rankings.
Scale score on PRBench Finance — November 12, 2026 snapshot
We mirror the published scale score view for PRBench Finance. Muse Spark 1.3 leads the public snapshot at 59.5%, followed by Muse Wonder 1.0 (58.7%) and Muse Spark 1.1\n (55.0%). We do not use these results to rank models overall.
42 modelsKnowledgeCurrentDisplay onlyUpdated November 12, 2026 snapshot
Scale score table (42 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Muse Spark 1.3Meta | 59.5% | Not reported | Closed |
| 2 | Muse Wonder 1.0Meta | 58.7% | Not reported | Unknown |
| 3 | Muse Spark 1.1\nMeta | 55.0% | Not reported | Unknown |
| 4 | claude-fable 5\nAnthropic | 53.9% | Not reported | Unknown |
| 5 | claude-opus-4-6 (Non-Thinking)Anthropic | 53.3% | Not reported | Unknown |
| 6 | Muse SparkMeta | 52.4% | Not reported | Closed |
| 7 | gpt-5OpenAI | 51.3% | Not reported | Unknown |
| 8 | gpt-5-proOpenAI | 51.1% | Not reported | Unknown |
| 9 | Fable 5.1\nAnthropic | 50.8% | Not reported | Unknown |
| 10 | gpt-5.6-sol (max)OpenAI | 50.5% | Not reported | Unknown |
| 11 | Gemini 4 ArgonGoogle | 49.9% | Not reported | Closed |
| 12 | Gemini 3.8 FlashGoogle | 49.5% | Not reported | Closed |
| 13 | o3-proOpenAI | 49.1% | Not reported | Closed |
| 14 | gpt-5.1-thinkingOpenAI | 48.0% | Not reported | Unknown |
| 15 | o3OpenAI | 47.7% | Not reported | Closed |
| 16 | GPT-6 AstraOpenAI | 47.5% | Not reported | Closed |
| 17 | Kimi K3p1 ProMoonshot AI | 47.5% | Not reported | Unknown |
| 18 | Deepseek V4p1 Flash 1017Deepseek | 47.1% | Not reported | Unknown |
| 19 | Claude Fable 5.2Anthropic | 47.1% | Not reported | Closed |
| 20 | Kimi K2.5Moonshot AI | 46.5% | Not reported | Open |
| 21 | GPT-5.2 ProOpenAI | 46.3% | Not reported | Closed |
| 22 | GPT-6.1 SolOpenAI | 46.2% | Not reported | Closed |
| 23 | Claude Opus 4.5 ThinkingAnthropic | 46.2% | Not reported | Closed |
| 24 | Minimax M3p1 FlashMiniMax | 45.6% | Not reported | Unknown |
| 25 | gpt-5.4 (High)OpenAI | 45.6% | Not reported | Unknown |
| 26 | GPT-OSS 120BOpenAI | 43.8% | Not reported | Open |
| 27 | Claude Sonnet 4.5Anthropic | 43.8% | Not reported | Closed |
| 28 | kimi-k2-thinkingMoonshot AI | 43.4% | Not reported | Unknown |
| 29 | Gemini 3.1 ProGoogle | 41.9% | Not reported | Closed |
| 30 | mistral-medium-latestMistral | 39.4% | Not reported | Unknown |
| 31 | o4-miniOpenAI | 39.2% | Not reported | Closed |
| 32 | gemini-3-pro-previewGoogle | 39.2% | Not reported | Unknown |
| 33 | qwen.qwen3-235b-a22b-2507-v1:0Alibaba | 39.1% | Not reported | Unknown |
| 34 | Gemini 2.5 ProGoogle | 38.9% | Not reported | Closed |
| 35 | Gemini 2.5 FlashGoogle | 38.4% | Not reported | Closed |
| 36 | kimi-k2-instructMoonshot AI | 38.3% | Not reported | Unknown |
| 37 | claude-opus-4-1-20250805Anthropic | 35.2% | Not reported | Unknown |
| 38 | deepseek-v3p1Deepseek | 35.1% | Not reported | Unknown |
| 39 | GPT-4.1OpenAI | 34.3% | Not reported | Closed |
| 40 | DeepSeek-R1-0528DeepSeek | 32.7% | Not reported | Open |
| 41 | GPT-4.1 miniOpenAI | 30.4% | Not reported | Closed |
| 42 | llama4-maverick-instruct-basicMeta | 22.4% | Not reported | Unknown |
How to read this leaderboard
Compare the published configurations as complete evaluation systems. The source can combine a base model, agent scaffold, tools, budget, and inference setting in each result.
Operator receipt: 42 sourced rows are currently displayable on this page; the leading published row is Muse Spark 1.3 at 59.5%.
Honest limit: This Scale table is display-only context, not benchmark provenance or a weighted model-only comparison.
How BenchLM shows PRBench Finance
BenchLM mirrors 42 published rows from Scale Labs’ public PRBench Finance leaderboard, captured on November 12, 2026 snapshot.
The table is display only. It is useful context for a published agent or model configuration, but it does not enter BenchLM’s overall or category rankings.
The published PRBench Finance snapshot places Muse Spark 1.3 first at 59.5%. The third row is 4.5 points behind. The broader top-10 range is 9.1 points, so many of the published results sit in a relatively narrow band.
42 models have been evaluated on PRBench Finance. The benchmark falls in the Knowledge category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. PRBench Finance is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About PRBench Finance
Year
2026
Tasks
42 published rows
Format
Published Scale leaderboard score
Difficulty
External agent and model evaluation
BenchLM mirrors 42 published rows from the PRBench Finance public table captured on November 12, 2026 snapshot.
Freshness and provenance
Version
PRBench Finance 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 PRBench Finance measure?
A Scale Labs public leaderboard mirrored as display-only reference data. It does not affect BenchLM rankings.
Which model leads the published PRBench Finance snapshot?
Muse Spark 1.3 currently leads the published PRBench Finance snapshot with 59.5% scale score. BenchLM shows this benchmark for display only and does not use it in overall rankings.
How many models are evaluated on PRBench Finance?
The November 12, 2026 snapshot snapshot contains 42 AI models.
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.