edgee-models

name	edgee-models
description	Update model metadata for the pi-edgee extension. Use when refreshing Edgee model lists, checking Edgee model availability, or syncing hardcoded model metadata with the Edgee API and models.dev.

Update Edgee models

Update src/edgee/models.ts from live Edgee and models.dev data, not guesses.

Default behavior

Take initiative:

1. If EDGEE_API_KEY is available, fetch https://edgee.io/v1/models to confirm API-visible models and their owned_by provider.
2. Edgee is a gateway, not a model host. It is NOT present on models.dev. Use models.dev upstream provider entries (e.g. openai) for pricing, context, output limits, modalities, and reasoning flags for each {provider}/{model} id.
3. Probe https://edgee.io/v1/chat/completions only when availability is unclear. Never print the API key.
4. Update src/edgee/models.ts.
5. Run model tests plus typecheck and lint.
6. Create a changeset for user-visible metadata changes.
7. Commit only relevant files when asked.

Do not push.

Sources of truth

Use these in order:

1. https://edgee.io/v1/models (with EDGEE_API_KEY) for the live model list and owned_by provider.
2. https://models.dev/api.json upstream provider entries (keyed by the provider prefix in the model id, e.g. openai/gpt-5.2 -> provider openai, model gpt-5.2) for pricing, context, output limits, modalities, and reasoning flags.
3. Edgee docs: https://www.edgee.ai/docs/api-reference/models
4. Existing hardcoded definitions for fields live sources do not expose.

Required checks

Run:

pnpm test -- src/edgee/models.test.ts
pnpm typecheck
pnpm lint

Field mapping

From models.dev (upstream provider entry, model id = part after the /):

• id -> id (keep the {provider}/{model} format from Edgee's list)
• modalities.input containing image -> input: ["text", "image"]
• reasoning -> reasoning
• limit.context -> contextWindow
• limit.output -> maxTokens
• cost.input -> cost.input
• cost.output -> cost.output
• cost.cache_read -> cost.cacheRead (default 0)
• cost.cache_write -> cost.cacheWrite (default 0)

Provider compatibility defaults:

compat: {
  maxTokensField: "max_tokens",
  supportsUsageInStreaming: true,
}

Runtime probes

Use these only when needed. Never echo EDGEE_API_KEY.

Check model list

curl -sS https://edgee.io/v1/models \
  -H "Authorization: Bearer $EDGEE_API_KEY" \
  -H 'accept: application/json'

Check chat completions

curl -sS https://edgee.io/v1/chat/completions \
  -H "Authorization: Bearer $EDGEE_API_KEY" \
  -H 'Content-Type: application/json' \
  -d @- <<'JSON'
{
  "model": "openai/gpt-5.2",
  "messages": [{"role": "user", "content": "Reply exactly ok"}],
  "max_tokens": 5
}
JSON

Decision rules

• Add a model when /v1/models or runtime confirms it is available.
• Keep hardcoded known models that are callable even if absent from GET /v1/models.
• Edgee compresses input tokens before the upstream provider sees them, so actual billed cost is lower than the models.dev list price. Cost metadata reflects upstream list price; do not adjust for compression.
• Leave cost at zero only when no pricing source exposes a price.
• Edgee is a gateway: never add Edgee itself to models.dev-based lookups.

Commit workflow

When asked to commit:

1. Run checks.
2. Check git status.
3. Stage only relevant files. Never use git add . or git add -A.
4. Use a concise conventional commit message.