BriefPulse Practical AI · Working notes on AI you can actually use. RSS · BriefPulse network
BriefPulse Practical AI

What changed in AI, what it is useful for, and what you can do with it.

15 September 2026

Guides / Practical AI

Guide

Check what a latest model alias points to before you ship

OpenRouter's catalogue lists entries that always follow the newest model in a family. This guide shows how to see what such an entry points to today, and when to call a fixed version name instead.

For Anyone whose script, app or agent calls a hosted model by name and needs to know which model actually answered.Checked 2026-09-15

What a latest alias is

OpenRouter's catalogue carries entries whose id begins with a tilde. Two examples are OpenAI: GPT Sol Latest (id ~openai/gpt-sol-latest) and DeepSeek: DeepSeek Flash Latest (id ~deepseek/deepseek-flash-latest).

Both listings describe themselves in the same terms: the model always redirects to the latest model in that family. The name you call stays the same; the model behind it does not.

How to see what it points at today

The Models API returns the whole catalogue as JSON from GET https://openrouter.ai/api/v1/models, and each alias entry in that response carries an alias_target object naming the model it currently counts as.

At the last check, on 15 September 2026, the list held 445 model entries, of which 18 were alias entries. ~openai/gpt-sol-latest resolved to OpenAI: GPT-5.6 Sol (slug openai/gpt-5.6-sol) and ~deepseek/deepseek-flash-latest resolved to DeepSeek: DeepSeek V4.1 Flash (slug deepseek/deepseek-v4.1-flash).

OpenRouter also documents a single-model lookup, GET /api/v1/model/{author}/{slug}, which resolves aliases automatically and returns 404 when the model does not exist. One request answers the question, so you never have to infer the target from a model page.

When to name a version instead

If a result has to be reproducible, such as an evaluation, a client deliverable or a support answer about which model produced an output, call the alias target's slug rather than the alias. A release then cannot move the model under you.

Prices come from the same catalogue. The GPT Sol Latest page shows 2 US dollars per million input tokens and 10 per million output tokens, with an override above 272,000 prompt tokens of 4 and 15. The DeepSeek Flash Latest page shows 0.15 per million input tokens and 0.60 per million output tokens.

The API states its pricing values in USD per token, request or unit, so a page that quotes per-million figures and an API that quotes per-token figures are saying the same thing. Re-check the figure on the day you quote it.

Two habits that keep you out of trouble

Keep the model id you tested next to the test result. If that id starts with a tilde, record the alias target slug as well, because that is the model you actually measured.

Check the catalogue on the day you re-run a cost estimate: an alias that has moved to a new model moves your input, output and cache prices with it.

What this guide does not cover

Only OpenRouter's catalogue is covered, and only what its public pages and API state. The pages do not say how quickly an alias follows a new release, and a catalogue entry is a listing rather than a quality claim. Alias targets and prices change, so re-check both before relying on either.

Links checked

  1. OpenRouter listing: OpenAI GPT Sol Latest — https://openrouter.ai/~openai/gpt-sol-latestchecked 2026-09-15, HTTP 200
  2. OpenRouter listing: DeepSeek Flash Latest — https://openrouter.ai/~deepseek/deepseek-flash-latestchecked 2026-09-15, HTTP 200
  3. OpenRouter Models API (alias targets and pricing, JSON) — https://openrouter.ai/api/v1/modelschecked 2026-09-15, HTTP 200
  4. OpenRouter Models API documentation (schema, pricing units, single-model lookup) — https://openrouter.ai/docs/overview/modelschecked 2026-09-15, HTTP 200