LLM Chatbot

GPT-5.4 and ChatGPT: OpenAI's 2026 Lineup

OpenAI's most capable model with standout computer use and Codex integration, though the $100/mo Pro plan sparked backlash for walling off professional features.

• $20/mo Plus, $100/mo Pro★★★★★ 5/5

GPT-5.4 and ChatGPT: OpenAI's 2026 Lineup

What It Is and Why You Should Care

GPT-5.4 is OpenAI's flagship AI model, and ChatGPT is the app you access it through. It's the Swiss Army knife of AI -- decent at everything, exceptional at a few things, and backed by the biggest name in the industry.

It's best for people who want one AI that can do it all: chat, generate images, control a browser, run autonomous coding tasks, and brainstorm creative ideas. If you want versatility and don't mind paying a premium for it, GPT-5.4 delivers.

  • Computer use -- controlling browsers, filling forms, automating web workflows
  • Creative brainstorming and divergent thinking where you want surprising ideas
  • Autonomous coding tasks via Codex (write tests, fix bugs, build features while you grab coffee)
  • General knowledge queries across a ton of domains
  • Having one subscription that covers chat, images, and coding

What Happened to OpenAI in 2026

OpenAI released GPT-5.4 on March 5, 2026, calling it their "most capable and efficient frontier model for professional work." They also launched GPT-5.3 Instant Mini as a faster, cheaper fallback model, and then on April 9 dropped a $100/mo Pro plan that immediately sparked backlash.

The short version: GPT-5.4 is excellent. The pricing strategy is... complicated.

Pricing: The Good, The Bad, and The $100

Chat Interface (chatgpt.com):

  • Free: GPT-5.3 Instant Mini (limited)
  • Plus ($20/mo): GPT-5.4 with usage limits, DALL-E 4, Codex access
  • Pro ($100/mo): Unlimited GPT-5.4, early feature access, higher rate limits

API Pricing:

  • GPT-5.4: $10/M input, $30/M output
  • GPT-5.3 Instant Mini: $0.15/M input, $0.60/M output

The $100/mo Pro plan is the controversy. OpenAI positioned it for "power users and professionals," but the reaction was sharp: many felt that features like unlimited GPT-5.4 access and Codex -- things professionals actually need -- were walled off behind a price that's 5x the Plus tier. For context, Claude Pro gives you both Sonnet and Opus for $20/mo.

If you're comparing API costs, GPT-5.4 is roughly 3x more expensive than Claude Sonnet 4.6 on input and 2x on output. Whether that's worth it depends entirely on your use case.

GPT-5.4: What It's Good At

GPT-5.4 is OpenAI's most capable model, and it shows in specific areas:

Computer use. This is GPT-5.4's standout feature. It can control a browser, navigate websites, fill forms, and execute multi-step workflows. In benchmarks, it beats humans on certain computer use tasks. For automating web-based workflows -- data entry, research, testing -- this is real.

Coding with Codex. The Codex integration means you can describe a coding task and have it executed in a sandboxed environment. It's not as tightly integrated into your IDE as Cursor, but for autonomous task execution (write tests, fix this bug, build this feature), it's powerful.

Creative work. GPT-5.4 generates more varied, creative output than Claude. If you want unexpected connections, novel ideas, or brainstorming that goes off-script, GPT-5.4 tends to produce more diverse suggestions.

General knowledge. Still the broadest knowledge base of any model. If you need information across many domains, GPT-5.4 has the edge.

Where GPT-5.4 Falls Short

Instruction following. Claude Sonnet 4.6 is better at following specific formatting and style instructions. GPT-5.4 tends to add its own flair, which is great for creativity but frustrating when you need precise output.

Honesty about uncertainty. GPT-5.4 is more likely to confidently give you a wrong answer than Claude. This matters a lot for research, analysis, and any work where accuracy is non-negotiable.

Context window. GPT-5.4's context window is 256K tokens -- large, but a quarter of Claude Sonnet's 1M. If you're working with large documents or codebases, Claude has a real advantage.

Cost. At every tier -- API, Plus, Pro -- OpenAI is more expensive than Anthropic for comparable work.

GPT-5.3 Instant Mini: The Unsung Hero

Don't overlook this one. GPT-5.3 Instant Mini is absurdly cheap on the API ($0.15/M input) and surprisingly capable for routine tasks. It's the model you should use for:

  • Classification tasks (sentiment analysis, categorization)
  • Simple extraction and formatting
  • Chatbot fallback when the heavy model is overkill
  • High-volume tasks where cost matters more than peak quality

Who Should Use What

ChatGPT Plus ($20/mo) if:

  • You want the most versatile all-in-one AI (chat + image gen + coding agent)
  • You value creative, diverse output over strict instruction following
  • Computer use is important for your workflow

Skip it if:

  • You primarily need long-context work (Claude's 1M window is better)
  • You need precise, structured output consistently
  • $100/mo for "real" access feels wrong (it should -- use Claude Pro instead)

The Bottom Line

GPT-5.4 is a great model trapped behind a pricing structure that's hard to love. At the Plus tier ($20/mo), you get genuine value -- especially with Codex and computer use. But the Pro tier's $100/mo price tag for features professionals need puts OpenAI at a disadvantage against Claude Pro's $20/mo offering. If you're API-first, GPT-5.3 Instant Mini is a steal. GPT-5.4 is good but overpriced relative to Claude Sonnet for most professional tasks.

See you next Tuesday. -James Anyway, if GPT-5.4 starts writing better sign-offs than me, we've got bigger problems.

Back to Tools