SRSHAW RANA

M-02 · MISSION LOG · 2026-09-16

The bake-off

34/34

IN 69 S

How I picked the model that does my typing.

8.1B
tokens, one week

One week of agents burned 8.1 billion tokens. Most of it was execution.

Spec
GPT-5.6 Luna
DeepSeek Flash
DeepSeek Pro
Verifier34 checks

Same spec. Same 34 checks. Only the model changed.

DeepSeek Flash69 s · 34/34
DeepSeek Pro79 s · 34/34
GPT-5.6 Luna189 s · 34/34

All three passed. The clock decided it.

2.7x
faster
$0.03
per run

Same score. Faster, cheaper, and no usage cap.

0 s0/34

Pick a lane. Watch the verifier run.

Claudeplans
DeepSeektypes
Claudereviews

DeepSeek now does the typing. Claude still reviews every result.

This lane built most of the site you're on.

SR-012 · The agent fleet

Open this star →Next mission · Paper to pipeline →
← MISSIONS

M-02 · 01 / 08