Add RodiumAi provider to Models.dev with 18 coding models

[rodiumai]

This PR adds the RodiumAi provider to models.dev so OpenCode users can select RodiumAi models with correct metadata (costs, context limits, tool calling, reasoning flags).

RodiumAi exposes an OpenAI-compatible API at https://api.rodiumai.io/v1. Model IDs in this PR match the slugs returned by GET /v1/models on the live gateway.

What's included

Path	Description
providers/rodiumai/provider.toml	Provider config (RODIUMAI_API_KEY, @ai-sdk/openai-compatible, API + docs URLs)
providers/rodiumai/logo.svg	Provider logo
providers/rodiumai/models/**/*.toml	18 curated models (code/agentic only: tool_call = true, text in/out)

No monolithic api.json edit — follows the current one TOML per model layout (same pattern as Groq, OpenRouter, etc.).

Models (18)

Smart routing (RodiumAi profiles)

ID	Role
auto	Smart routing (default)
basic	Cheapest tier
fast	Low-latency tier
pro	Balanced / large context
max	Frontier tier

Vendor slugs (passthrough to upstream)

Vendor	Models
Anthropic	anthropic/claude-opus-4-7, anthropic/claude-sonnet-4-6, anthropic/claude-haiku-4-5-20251001
OpenAI	openai/gpt-5.5, openai/gpt-5.4, openai/gpt-5.4-mini, openai/o3
Google	google/gemini-3.5-flash, google/gemini-3.1-pro-preview, google/gemini-2.5-flash, google/gemini-2.5-pro
DeepSeek	deepseek/deepseek-v4-pro, deepseek/deepseek-v4-flash

Scope: coding assistants only — no image/video/embedding models. Smart-profile pricing is documented in TOML; vendor pricing is sourced from RodiumAi's public catalog scrapes (USD per 1M tokens, optional cache read/write where applicable).

Provider configuration

name = "RodiumAi"
env = ["RODIUMAI_API_KEY"]
npm = "@ai-sdk/openai-compatible"
api = "https://api.rodiumai.io/v1"
doc = "https://rodiumai.io/docs"

How users enable it (after merge)

export RODIUMAI_API_KEY=sk-...
opencode auth login
opencode --model rodiumai/pro
opencode --model rodiumai/google/gemini-2.5-flash

Maintenance

Manifest and generator live in the RodiumAi monorepo (not required for this PR to merge):

• Curated list: rodiumai_docs/opencode/rodiumai-curated-models.json
• Regenerate TOML: cd rodiumai_nestjs && npm run models-dev:generate

Future model additions should keep slug parity with GET https://api.rodiumai.io/v1/models.

Test plan

• bun run validate (or project-equivalent) passes for providers/rodiumai/**
• Build packages/web and confirm rodiumai appears in _api.json
• OpenCode: opencode --model rodiumai/auto completes a tool-capable chat turn
• OpenCode: at least one vendor slug (e.g. rodiumai/anthropic/claude-sonnet-4-6) works end-to-end
• Costs/limits in TOML are reasonable vs. provider docs (spot-check pro, gpt-5.5, gemini-2.5-flash)

Links

• API: https://api.rodiumai.io/v1
• Docs: https://rodiumai.io/docs
• RodiumAi fork (reference): https://github.com/Rodium-AI/models.dev