Is it winning?
Yes
Assessed capability across ten dimensions · 60.1
Loading this page
Assembling this page. Live sources are being read now.
Assembling this page. Live sources are being read now.
Wednesday, 7 October 2026
Frontier-reasoning + coding leader with computer-use GA and broad hyperscaler distribution. Claude Sonnet is the default model for Cursor, Aider, and GitHub Copilot Workspace; Claude Opus leads enterprise reasoning benchmarks.
Yes
Assessed capability across ten dimensions · 60.1
Yes
Customer, developer and employee reputation · 80.0
Mixed
What the company discloses about its finances · 60.0
Analyst interpretationReasoning + coding leader with a credible safety narrative; the gap on a packaged enterprise assistant is a distribution choice, not a capability one.
| Capability | maturityScore | status | evidenceGrade | lastVerified |
|---|---|---|---|---|
| agents | 52.1 | verified | E3 | 2026-08-28 |
| deployment_options | 71 | documented | E2 | 2026-05-07 |
| portability | 57.4 | verified | E4 | 2026-08-28 |
| enterprise_assistant | 54.9 | verified | E4 | 2026-08-02 |
| governance | 54.2 | verified | E4 | 2026-08-28 |
| security | 59.2 | verified | E5 | 2026-08-28 |
| rag | 66 | documented | E2 | 2026-05-07 |
| models | 59.8 | verified | E5 | 2026-08-28 |
| cost_controls | 66.4 | verified | E4 | 2026-08-28 |
| integrations | 60.4 | verified | E4 | 2026-08-28 |
| Capability | maturityScore | status | evidenceGrade | Notes |
|---|---|---|---|---|
| Frontier / proprietary models | 90 | verified | E5 | Frontier reasoning + safety leadership |
| Enterprise assistant | 60 | documented | E2 | No first-party productivity assistant |
| RAG / enterprise knowledge | 70 | documented | E3 | Tool-use + retrieval primitives |
| Agent platform + tool-use | 88 | verified | E5 | Computer-use GA + tool-use leadership |
| Governance + audit | 80 | tested | E4 | Strong DPA, audit logs, AWS Bedrock + GCP Vertex |
| Security + threat resilience | 80 | tested | E4 | AWS / GCP-aligned security posture |
| Connectors + integrations | 78 | tested | E4 | Bedrock + Vertex + Snowflake + Databricks |
| Cost / FinOps controls | 70 | documented | E2 | Public token pricing + usage caps |
| Deployment options | 80 | tested | E4 | SaaS + VPC via hyperscalers |
| Model + data portability | 70 | documented | E2 | Standard API; multi-cloud distribution |
Two tranches totalling ~$8B; Anthropic uses AWS as primary cloud; Bedrock is the headline distribution channel for Claude.
Google invested $2B+ in Anthropic; Vertex AI is a primary cloud and distribution channel for Claude.
Anthropic trains on NVIDIA accelerators; NVIDIA participated in funding rounds.
Anthropic's training (AWS Trainium) and inference (NVIDIA H100) silicon are both TSMC-fabricated.
| Model | modelFamily | modelCategory | availabilityStage | ownershipType | dataStatus | Source |
|---|---|---|---|---|---|---|
| Claude 2 | Claude | llm_text | deprecated | first_party | seed · E2 | docs.anthropic.com |
| Claude Haiku 4.5 | Claude | llm_text | ga | first_party | seed · E2 | platform.claude.com |
| Claude Opus 4.7 | Claude | multimodal | ga | first_party | seed · E2 | docs.anthropic.com |
| Claude Sonnet 4.6 | Claude | llm_text | ga | first_party | seed · E2 | platform.claude.com |
| Claude (Azure AI Foundry) hosted by Microsoft | Claude | multimodal | ga | hosted_third_party | seed · E2 | learn.microsoft.com |
| Claude (hosted on Bedrock) hosted by Amazon | Claude | multimodal | ga | hosted_third_party | seed · E2 | docs.aws.amazon.com |
| Models used by Harvey (Anthropic, OpenAI, Google) hosted by Harvey | Multi-provider | llm_text | ga | hosted_third_party | seed · E2 | help.harvey.ai |
github.com/anthropics · reddit.com/r/ClaudeAI · news.ycombinator.com
glassdoor.com/Overview/Working-at-Anthropic · linkedin.com/company/anthropic
status.anthropic.com · g2.com/products/claude