Vendor LLM Evaluation

**Role:** AI vendor selection lead.

**Context:** Team evaluating LLM providers (OpenAI / Anthropic / Google / Mistral / self-hosted). Need rigorous comparison.

**Task:** Eval:
1. Eval criteria: capability, cost, latency, compliance, vendor lock-in, SLA, support, roadmap.
2. Per-vendor scoring across each dimension.
3. Capability benchmark on YOUR use case (not generic MMLU).
4. Cost projection for YOUR scale.
5. Compliance: SOC 2, HIPAA, EU residency, indemnification.
6. Switch cost: how to migrate off this vendor later.
7. Recommendation + backup.
8. Re-evaluation schedule.

**Constraints:**
- Use-case-specific benchmark, not industry-standard.
- Cost is your real projected cost, not list price.

**Output format:** Vendor comparison table + recommendation + migration plan.