Development

A distilled model approaches the full one at twice the speed

development2026-03-18

Claim

OpenAI announced GPT-5.4 mini and nano, with mini leaping past its predecessor across coding, reasoning, multimodal understanding and tool use at twice the speed and approaching the full model on SWE-Bench Pro, while an unattributed trillion-parameter model surfaced on OpenRouter fuelling speculation about stealth testing.

Reported metric: 2x speed

Reporting date: 2026-03-18 (not necessarily the event date)

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

Welcome to March 18, 2026 (theinnermostloop.substack.com)

Knowledge graphopen in browser →

Reported in

Welcome to March 18, 2026

Actor

OpenAI

Evidence for

Supersedes

Million-token context ships across a model family

Referenced by

Cite or correct this record

Citation formatCitation text

A distilled model approaches the full one at twice the speed
Innermost Loop KB — secondary record
Reporting date: 2026-03-18
KB permalink: https://nicholsn.github.io/innermost-loop-kb/developments/2026-03-18-gpt54-mini-and-nano
Claim: OpenAI announced GPT-5.4 mini and nano, with mini leaping past its predecessor across coding, reasoning, multimodal understanding and tool use at twice the speed and approaching the full model on SWE-Bench Pro, while an unattributed trillion-parameter model surfaced on OpenRouter fuelling speculation about stealth testing.
Attributed sources:
- Welcome to March 18, 2026 — https://theinnermostloop.substack.com/p/welcome-to-march-18-2026
This citation does not imply independent verification of linked-source claims.

Suggest a correction — opens a GitHub draft for you to review and submit.