Summary
Two wrongs in the same place.
A call on a subscription is costed at zero. ChatGPT (Codex), a Copilot seat, and a
MiniMax coding plan are not billed per token, so no per-token figure describes what
a call cost. LiteLLM files those models at a price of zero, the estimate reads that
as a real number, and the session reports $0.00 -- indistinguishable from a free
model, and accumulating into a session total that means nothing.
The subscription prices Raven displays were written into the source without a
source.
Steps to reproduce
- Sign in to
openai-codex and select one of its models.
- Send any message.
- Look at the cost reported for the turn and for the session.
Expected behavior
"No per-token price" is a distinct answer from zero and is not accumulated into a
session total. How a provider bills is declared rather than inferred from whether it
uses OAuth -- Vertex is OAuth and metered.
Actual behavior
The turn and the session both report $0.00 on a paid plan.
Environment
OS: macOS 26.5 (arm64)
Shell: zsh
Python: 3.12.12
Node: v24.14.0
Raven: 3087174 (origin/main)
Install: source checkout, uv sync
litellm: 1.85.0
Summary
Two wrongs in the same place.
A call on a subscription is costed at zero. ChatGPT (Codex), a Copilot seat, and a
MiniMax coding plan are not billed per token, so no per-token figure describes what
a call cost. LiteLLM files those models at a price of zero, the estimate reads that
as a real number, and the session reports
$0.00-- indistinguishable from a freemodel, and accumulating into a session total that means nothing.
The subscription prices Raven displays were written into the source without a
source.
Steps to reproduce
openai-codexand select one of its models.Expected behavior
"No per-token price" is a distinct answer from zero and is not accumulated into a
session total. How a provider bills is declared rather than inferred from whether it
uses OAuth -- Vertex is OAuth and metered.
Actual behavior
The turn and the session both report
$0.00on a paid plan.Environment