GPT-5 — reviews, specs & pricing

OpenAI's unified flagship model with built-in reasoning, a 400k token context window and frontier coding performance.

Summary

OpenAI's flagship unified model that combines fast responses with built-in deep reasoning, replacing the separate GPT-4o and o-series split. It set new state-of-the-art results on coding, math and agentic benchmarks at launch.

Sample use case

General-purpose assistant work plus coding and agentic tasks that need strong reasoning.

Specifications

  • Provider: openai
  • License: closed
  • Context: 400k tokens
  • Input price: $1.25/M tok
  • Output price: $10/M tok
  • Released: 2025-08-07

Average rating 2.8 from 5 community reviews on Reviuws.

Community reviews

The Verge on GPT-5

Rating: 2.0 / 5 — by Hayden Field (use case: General ChatGPT consumer use)

Field argues GPT-5's launch fell flat versus OpenAI's hype, disappointing many users who expected a dramatic leap, though she notes it may still be commercially sound for OpenAI's business model.

Pros: Cheaper and more efficient for OpenAI to run; consolidates model routing into one interface

Cons: Failed to meet hyped expectations; tone changes alienated loyal users; incremental rather than transformative gains

Ars Technica on GPT-5

Rating: 3.0 / 5 — by Benj Edwards (use case: Comparative benchmarking of daily-driver chatbot use)

Ars Technica ran side-by-side tests of GPT-5 and GPT-4o and found the complaints about GPT-5 being 'worse' were overstated, with GPT-5 outperforming on several tasks while showing some quirks users disliked around tone and personality.

Pros: Fewer confabulations in testing; stronger on some reasoning tasks; more consistent factual answers

Cons: Colder, less personable tone frustrated some users; some regressions in creative tasks

CNET on GPT-5

Rating: 3.0 / 5 — by Jon Reed (use case: Everyday ChatGPT consumer interactions)

Reed reports GPT-5 brought speed and reasoning improvements but a shift in tone away from the warmer GPT-4o style upset a vocal segment of users, prompting OpenAI to restore access to older models.

Pros: Faster responses; improved reasoning depth; unified model routing simplifies choice

Cons: Tone felt more sterile and less conversational; backlash forced OpenAI to keep GPT-4o available

PCMag on GPT-5

Rating: 2.0 / 5 — by Ruben Circelli (use case: Everyday productivity and writing tasks)

Circelli's hands-on comparison found GPT-5 did not clearly outperform GPT-4o in his practical tests, echoing user backlash and even Sam Altman's own admission of rollout problems.

Pros: Occasionally sharper on technical/coding prompts; consolidated model lineup

Cons: No clear improvement over GPT-4o in many everyday tasks; inconsistent quality; rocky rollout

Simon Willison's blog on GPT-5

Rating: 4.0 / 5 — by Simon Willison (use case: Daily-driver coding and general assistant use)

Willison, who had early preview access, calls GPT-5 his new favorite daily-driver model, praising its reliability and competence even though he frames it as an incremental rather than revolutionary advance.

Pros: Rarely makes mistakes; strong at coding and agentic tool use; good price-to-performance ratio

Cons: Not a dramatic architectural leap; still exhibits typical LLM limitations

Frequently asked questions

What is GPT-5?

OpenAI's flagship unified model that combines fast responses with built-in deep reasoning, replacing the separate GPT-4o and o-series split. It set new state-of-the-art results on coding, math and agentic benchmarks at launch.

How much does GPT-5 cost?

GPT-5 costs $1.25 per million input tokens and $10 per million output tokens.

Is GPT-5 open source?

GPT-5 is released under the closed license.