kwj.ai · acquisition inquiries from >$999view prospectus →
The Domesday Book ofKWJ · AI

OpenAI · The Long-Context Stretch

GPT-4.5

The last non-reasoning OpenAI flagship. Better at vibes than reasoning. Quietly deprecated in favor of GPT-5.

By C.W. Jameson · Published 19 May 2026 · Last reviewed 19 May 2026

GPT-4.5 was OpenAI's attempt at scaling pretraining one more turn. The result was a model that wrote with palpable warmth, refused less aggressively than 4o, and lost the reasoning benchmarks to its own little sibling o3. It was deprecated within a year. Worth knowing because legacy production code still depends on it and the model's 'voice' shows up in fine-tuned derivatives.

Field signature

Notably warm prose. Lower refusal rate than 4o. Will give an opinion if asked directly.

Specifications

Released2025-02
Context window128,000 tokens
PricingDeprecated
Modalitiestext · image · audio
LicenseCommercial API only, deprecated
EraThe Long-Context Stretch

Strengths

  • Voice and prose quality
  • Conversation comfort

Weaknesses

  • Reasoning benchmarks
  • Officially deprecated

Authentication markers

The fingerprints by which GPT-4.5 can be identified from its output alone.

TellMeaning
Em-dash use noticeably higher than GPT-4o.GPT-4.5 lineage.

Notable works

  • Powered late-2025 ChatGPT consumer experience before GPT-5 cutover

Market position

Deprecated

Partner offer

OpenAI's API surface remains the broadest commercial offering.

Try OpenAI →

Affiliate link — see disclosure.

Primary sources

  1. [1] OpenAI: Models

From the Almanac shop

The Operator's Compendium

Every agent harness, every routing pattern, every cost trick. 90-page PDF.

$29Coming soon

Back to the directory