Twelve dimensions across seven providers. Re-graded as announcements land, with the market's reaction shown alongside.
Raw model intelligence is converging at the top. The real contest is orchestration, context, distribution, and trust. These are editorial grades, a considered read of the public record rather than a benchmark, held to a transparent rubric and re-graded live as announcements land. Baseline as of June 1, 2026.
Intelligence wired into the surfaces a billion people already use.
#2AI metered into the world's enterprise workflows.
#3The assistant most people mean when they say 'AI'.
| Dimension | #1Google7.8 | #2Microsoft7.4 | #3OpenAI7.2 | #4Anthropic7.0 | #5Meta6.1 | #6Apple5.8 | #7xAI4.8 |
|---|---|---|---|---|---|---|---|
Assistant Intelligence Raw capability | 9 | 8 | 9 | 9 | 7 | 6▲+2 | 8 |
Agent Capability Can it actually do work? | 7 | 7 | 8 | 9 | 5 | 5▲+1 | 5 |
Conversational UX Natural interaction quality | 8 | 7 | 9 | 6 | 7 | 6▲+2 | 7 |
Cross-App Actions Real orchestration | 7 | 8 | 5 | 7 | 4 | 6▲+1 | 3 |
Personal Context Memory and awareness | 8 | 8 | 7 | 4 | 6 | 8▲+1 | 4 |
Third-Party Integrations Ecosystem reach | 8▲+1 | 8 | 7 | 8 | 7 | 4 | 4 |
Developer APIs Platform leverage | 8 | 8 | 9 | 9 | 7 | 5 | 6 |
Model Strategy Own the stack vs depend on partners | 9 | 6 | 9 | 8 | 8 | 4 | 7 |
Agent Platform Potential Future control-plane potential | 8 | 8 | 8 | 8 | 6 | 5 | 5 |
On-Device AI Privacy and latency | 7 | 6 | 2 | 1 | 7 | 9 | 2 |
Enterprise Readiness Business adoption | 8 | 9 | 7 | 8 | 4 | 7 | 3 |
MyAGI Alignment How close to an agent operating system | 7 | 6 | 6 | 7 | 5 | 5▲+1 | 4 |
Overall Equal-weight average | 7.8 | 7.4 | 7.2 | 7.0 | 6.1 | 5.8 | 4.8 |
Our north-star lens: how close is this provider to a true agent operating system? Earned (not asserted) capability, durable institutional memory, cross-app orchestration, and revocable, governable execution. The control plane for the agent era.
Current leader: Google (7.8 overall)