GPT 5.6 Sol Scores Revealed on Financial Benchmark
maithra_raghu · x · 2026-07-14
This post updates the results of GPT 5.6 Sol on the FrontierFinance benchmark, described by the author as one of the "hardest and largest" public financial intelligence evaluations.
Main Results
- GPT 5.6 Sol: 46.8%
- Fable 5: 49.2%, higher quality
- Opus 4.8: 45%
- Samaya’s System (light): 50.8%, lowest cost while remaining on the Pareto frontier of quality/cost
- Sol's advantage: Single-query cost is lower than Fable
Author's Analysis
- Different models show varying performance in "financial data extraction" and "financial analysis capabilities"
- The author also provided detailed case studies explaining which types of questions are harder for specific models
More from Models
- Rumor claims Kimi faked performance by serving Claude; DeepSeek new model surprises in evals — realsohamparekh · 2026-09-11
- GPT-5.6 writes well but is instantly forgettable, user complains — BasedRaddka · 2026-09-11
- Opus Refuses Protein Research Codebase Over 'Safety' Concerns, Dev Considers Rolling His Own — josephdviviano · 2026-09-11
- User Hails Unconfirmed 'DeepSeek 4.1 Flash' as an Inflection Point in LLMs — himanshustwts · 2026-09-11
- Terminal Bench v4: GLM-5.3 Leads at 41.9%, Kimi-K3 Underwhelms at 12.6% — Ok_Warning2146 · 2026-09-11
- GPT-6 Astra beats Factorio with enemies in 44 in-game hours at ~$4,500 API cost — liminal_bardo · 2026-09-11