Dated card / reviewed 17 Aug 2026

General-purpose model providers

A shortlist is not a winner. Run the same real evaluation set under the plan, region, data controls and tools you will actually use.

ProviderCapability familyUse this card forPrimary source
OpenAI
Observed
General reasoning, multimodal, coding, tools and agentic workflows.Evaluate by job across product and API; availability varies.Official documentation ↗
Anthropic
Observed
Reasoning, long-form work, coding and tool-enabled workflows.Evaluate instruction following, context use and tool boundaries.Official documentation ↗
Google
Observed
Gemini consumer, developer and workspace ecosystem plus multimodal models.Verify capability in the target region, product tier and API.Official documentation ↗
Microsoft
Observed
Workplace and cloud AI shaped by organisational identity, data and permissions.Evaluate integration and inherited access—not chat quality alone.Official documentation ↗

Prices, context limits and model names are intentionally omitted here unless they are reviewed on the date of use. Capability descriptions are qualitative and not endorsements.