What the ChatGPT 5? Is it free, what’s new, how you can get it…all questions answered

GPT-5 is being positioned as a step ahead in OpenAI’s long-term pursuit of synthetic basic intelligence. Whereas the mannequin doesn’t meet AGI benchmarks as outlined in OpenAI’s personal constitution, it’s described internally as typically clever. It doesn’t have the flexibility to study after deployment.

The corporate claims GPT-5 is extra correct and environment friendly, with decreased hallucination charges in comparison with earlier variations. OpenAI’s CEO Sam Altman stated the leap from GPT-4 to GPT-5 is corresponding to the show improve from early iPhones to Retina shows.

Two new mannequin variants have been launched: GPT-5-mini, a light-weight model, and GPT-5-nano, a sooner and cheaper mannequin accessible solely by way of API. Free customers can have entry to GPT-5 and GPT-5-mini. Paid tiers unlock increased utilization limits, with the $200/month Professional tier providing GPT-5-pro, GPT-5-thinking, and limitless entry.

The ChatGPT interface now auto-selects essentially the most appropriate mannequin primarily based on question complexity and consumer plan. Legacy fashions stay accessible for Professional customers.

On the API entrance, GPT-5 is priced at $1.25 per million enter tokens and $10 per million output tokens. GPT-5-mini is priced at $0.25 per million enter and $2 per million output tokens. GPT-5-nano is priced at $0.05 per million enter and $0.40 per million output tokens. The nano variant is now priced decrease than Google’s Gemini 2.5 Flash.

From subsequent week, Professional customers will be capable of combine Gmail, Google Calendar and Contacts with ChatGPT. Different tiers will obtain entry later. ChatGPT will reference these accounts routinely when related, with out guide choice.

Customers can now choose from 4 preset personalities: Cynic, Robotic, Listener, and Nerd. These modes are additionally deliberate for integration with Superior Voice Mode. The chat color can be customised.

The API permits toggling between detailed and direct responses. GPT-5 incorporates a 256,000-token context window, up from 200,000 within the earlier o3 mannequin. This allows higher retention throughout lengthy conversations, code and paperwork.

GPT-5 outperforms earlier fashions in coding duties. Benchmarks embrace 74.9 % on SWE-Bench Verified, 55 % on SWE-Lancer utilizing GPT-5-thinking, and 88 % on Aider Polyglot. These cowl bug fixing, multi-language help and freelance-style duties.

Throughout a demo, OpenAI’s group instructed GPT-5 to construct an interactive language-learning app. The system generated the requested options and UI inside a minute. OpenAI claims the mannequin can deal with advanced job chains and power use, together with browser and API-based calls.

In well being benchmarks, GPT-5-thinking scored 25.5 % on HealthBench Arduous, in comparison with o3’s 31.6 %. OpenAI states that its new mannequin performs higher throughout health-related duties, with outputs reviewed by medical professionals.

The corporate stories that hallucination charges have declined. GPT-5-thinking reveals a 65 % decrease fee in comparison with o3, and 26 % decrease than GPT-4o. Security updates embrace decreased possibilities of the mannequin producing misleading, hacked or unsafe outputs. Twin-use prompts are dealt with with what OpenAI calls “protected completions”, providing responses inside managed boundaries. Over 5,000 hours of purple teaming and exterior testing had been carried out to enhance robustness.

The system is designed to fail safely when confronted with unsolvable prompts. This consists of refusing duties past its capabilities with out producing deceptive outputs.

OpenAI stories 700 million weekly ChatGPT customers, 5 million paying enterprise customers, and 4 million builders on the API.

What’s new in GPT-5: Key adjustments at a look:

• Accessible to All Customers: GPT-5 is being rolled out to each free and paying ChatGPT customers.
• New Mannequin Variants:
  • GPT-5-mini (light-weight)
  • GPT-5-nano (API solely, sooner and cheaper)
  • GPT-5-pro and GPT-5-thinking (for Professional-tier customers)
• Computerized Mannequin Choice: The chat interface now auto-selects the mannequin primarily based on the complexity of the question and consumer tier. Guide switching is not required.
• Professional-Tier Advantages:
  • Limitless GPT-5 entry
  • Entry to GPT-5-pro and GPT-5-thinking
  • Legacy fashions nonetheless accessible
• Google Integration: Professional customers can quickly join Gmail, Google Contacts, and Google Calendar to ChatGPT. Different tiers will get this later.
• Customisation Options:
  • Customers can select chat colors
  • 4 built-in persona modes: Cynic, Robotic, Listener, Nerd
  • Will probably be built-in with Superior Voice Mode later
• Efficiency Boosts:
  • Context window elevated to 256,000 tokens (up from 200,000 in earlier o3 mannequin)
  • Decrease hallucination fee than GPT-4o and o3
  • Improved efficiency in coding benchmarks like SWE-Bench, SWE-Lancer, and Aider Polyglot
  • Higher at agentic duties and power execution (e.g., API and browser utilization)
• Developer Pricing:
  • GPT-5: $1.25/1M enter tokens, $10/1M output tokens
  • GPT-5-mini: $0.25/1M enter, $2/1M output
  • GPT-5-nano: $0.05/1M enter, $0.40/1M output
• Well being-related Efficiency:
  • GPT-5-thinking outperforms o3 in well being benchmarks
  • Claimed to be OpenAI’s greatest mannequin for medical questions
  • Evaluated by physicians; designed to fail gracefully on unsolvable duties
• Security Enhancements:
  • 26% discount in hallucinations over GPT-4o
  • 65% discount over o3 for GPT-5-thinking
  • Safer completions for dual-use prompts
  • 5,000+ hours of purple teaming and exterior security testing
  • Decreased chance of misleading or unsafe behaviour