Databricks says Claude Opus 5 tops its coding benchmark and ships in AI Gateway
thesaraharminta · x · 2026-07-25
Databricks says Claude Opus 5 is now available in its AI Gateway, with governed access to data already in the Lakehouse.
An internal Databricks evaluation claims Opus 5 is a clear step up from Opus 4.8:
- Coding: top score on their internal coding bench, beating even Fable 5 while costing about the same as Opus 4.8.
- OfficeQA Pro: 60.9 vs 48.12 for Opus 4.8, and roughly on par with Fable 5.
- Enterprise agentic search: ranked #4 on the bench, but still clearly ahead of Opus 4.8.
Databricks says the model brings better agentic coding, knowledge-work performance, and long-horizon reasoning, with stronger performance per token and lower cost at production scale.
More from Companies & People
- Chrome extension usage jumps 8× in five months as founder hunts for growth help — josh_bickett · 2026-07-25
- AI Security Forum still has open seats with speakers from Anthropic and ARIA — joshua_saxe · 2026-07-25
- A Venice Beach startup photo set turns “trillion-dollar consumer health” into a meme — arthurcolle · 2026-07-25
- Packy McCormick: $8 trillion of market cap now backs open-weight models — ctjlewis · 2026-07-25
- Archetype says its Newton model makes sense of industrial sensor data — plopesresearch · 2026-07-25
- Internal agents hit the company wall when they need outside systems — Just_Acanthisitta673 · 2026-07-25