Skip to content
Back
PRICING BRIEFINGDaily · Oct 30, 2026
BBaseten
Baseten
PACKAGING CHANGE

Baseten adds Base Labs joey-1 model to Model API rate table at $0.01 input

Baseten added a new row to its Model API rate table for Base Labs' joey-1 model, priced at $0.01 input with no cached-input or output rate shown and a "Try Model API" link, same as its other listings.

Baseten’s pricing page · Change Detected

WHY IT MATTERS

A model-hosting platform that changes its pricing page roughly every 11 days, with three new models this month, is competing on catalog breadth and price per token as much as on any single rate. Watch whether this pace holds into November or whether it is followed by a broader rate cut across the existing catalog, the way Baseten already cut Reflection Beam's rates once this quarter.

BY THE NUMBERS

  • 11th tracked pricing event, 1 day after the previous one (Oct 29, 2026)
  • Third new Model API add-on this month (Oct 16, 17 and 30)
  • Averages a pricing change every 11 days since tracking began Jul 20, 2026
  • Tied with Vercel for the most frequent company shipping new add-ons in the last 180 days

THEIR RATIONALE

Baseten keeps expanding its Model API catalog rather than repricing what is already listed. This is its 11th tracked pricing event, and it landed just one day after elephant-1 and mammoth-1 had their input pricing revealed, continuing an unusually fast cadence that suggests catalog growth is treated as a default, ongoing motion rather than a periodic update.

TRACK RECORD10 changes since Jul 20, 202610 in the last 12 monthsa change every ~11 daysprevious change Oct 29, 2026 (1 day ago)

THE SIGNAL

This is Baseten's third new-model add-on this month, after Reflection Beam and Meta Muse Spark Open joined the lineup on Oct 17 and Nemotron 3.5 Super VL was added on Oct 16. Across the index, new add-ons made up 125 of the 2,350 pricing changes tracked in the last 180 days across 108 companies, and Baseten itself is tied with Vercel for the most frequent company doing it in that window. At $0.01 input, joey-1 is priced below elephant-1 ($0.035) and mammoth-1 ($0.10), and below the $0.20 Nemotron 3.5 Super VL added on Oct 16.

NEW ADD-ONS108 companies in the last 6 months5.3% of 2,350 tracked changes3× on Baseten's pageVercel most often (4)

IN CONTEXT

Baseten keeps adding to its Model API catalog at a pace few peers match: joey-1 lands just a day after elephant-1 and mammoth-1 had their input pricing revealed, and it is Baseten's third new-model add-on this month, after Reflection Beam and Meta Muse Spark Open joined the lineup on Oct 17 and Nemotron 3.5 Super VL was added on Oct 16. Across the index, new add-ons accounted for 125 of the 2,350 pricing changes tracked in the last 180 days across 108 companies, and Baseten itself is tied with Vercel for the most frequent company doing it in that window. At $0.01 input, joey-1 is priced below elephant-1 ($0.035) and mammoth-1 ($0.10), listed the day before.

WHAT PEOPLE ARE SAYING

@DanteMMartintwitter

Baseten's joey-1 is a 2B open-weight model that doesn't write — it picks an answer and returns a probability. $0.01 per 1M tokens, ~20ms a request. Most AI spend goes to prose nobody reads. That's the fix. https://www.marktechpost.com/2026/10/29/baseten-joey-1-2b-decision-model/

View original
@Apurv35904761twitter

don't send yes/no routing to a giant llm. joey-1 (baseten): 2b decision model from qwen3.5-2b. state + typed questions → calibrated probs. $0.01/1m input, ~20ms p50 (baseten). open weights on hf. tiny model for the boring decisions. https://www.baseten.co/blog/decisions-for-a-penny-introducing-joey-1/

View original
@joeybesgentwitter

crushing the model-naming game

View original