Llm Advisor
Compare live LLM and VLM benchmark scores, pricing, and picks while choosing models for your agent product or API stack.
Overview
io.github.Daichi-Kudo/llm-advisor is an MCP server for the Build phase that delivers real-time LLM/VLM benchmarks, pricing, and recommendations across 336+ models from 5 sources.
What is this MCP server?
- 336+ LLM/VLM models with benchmarks and pricing
- Data aggregated from 5 sources in real time
- Stdio MCP npm package llm-advisor-mcp (v0.4.2)
- Recommendations oriented to coding-agent workflows
- 5 benchmark/pricing sources
- Package version 0.4.2 (llm-advisor-mcp)
What problem does it solve?
Model catalogs change weekly and spreadsheet comparisons of price versus benchmark scores slow down agent and API design.
Who is it for?
Indie builders picking inference providers for agents, RAG, or VLMs who want consolidated benchmark and pricing signals in-chat.
Skip if: Teams that already locked a single vendor contract and do not revisit model routing.
What do I get? / Deliverables
Your coding agent can pull up-to-date model rankings, costs, and advice through MCP while you wire providers and fallbacks.
- Agent-accessible model recommendations and benchmark snapshots
- Pricing context across 336+ models from 5 sources
Recommended MCP Servers
Journey fit
Model selection and cost tradeoffs happen while you are assembling agent tooling and inference dependencies in the build phase. Agent-tooling is the canonical shelf because the server feeds model-choice decisions that directly affect prompts, routers, and fallbacks.
How it compares
Live model-intelligence MCP, not a hosting gateway or fine-tuning skill.
Common Questions / FAQ
Who is io.github.Daichi-Kudo/llm-advisor for?
Solo builders and agent authors who need fast, sourced comparisons of LLM and VLM options while coding integrations.
When should I use io.github.Daichi-Kudo/llm-advisor?
Use it during build and early validate when selecting default models, backup tiers, or vision stacks before you commit to SDK wiring.
How do I add io.github.Daichi-Kudo/llm-advisor to my agent?
Add stdio MCP with npm package llm-advisor-mcp per your client’s config; no secrets are listed in the published server schema.