Tendril

Tendril · Builders · Model Families

AI model families: GPT-5 and what's new

Understand what makes GPT-5 different from GPT-4 and earlier OpenAI models.

40 min · Reviewed 2026

The big idea

GPT-5 is OpenAI's flagship model. It's smarter at reasoning, handles longer documents, and is better at code. Knowing what version you're using matters because answers differ.

Some examples

Use GPT-5 for hard reasoning tasks
Notice GPT-5's better code than GPT-4
Compare GPT-5 vs GPT-4 on the same prompt
Check which model your tool actually uses

Try it!

Find a hard math or logic question. Ask it to GPT-4 and GPT-5 (if you have access). Note where the answers differ. Even if both right, the explanations vary.

AI and GPT-5: OpenAI's Latest Big Brain

The big idea

GPT-5 is OpenAI's frontier model. It can read images, write code, do math, and reason through hard problems. ChatGPT Plus uses GPT-5 by default for paying users.

Some examples

Upload a photo and ask GPT-5 to identify what's in it.
Use GPT-5 for coding when Sonnet gets stuck.
GPT-5 is great at structured outputs (JSON, tables).
It's slower than smaller GPTs but more accurate on hard stuff.

Try it!

Try the same hard question on GPT-5 and a smaller model (Haiku or 4o-mini). Compare which one nails it.

AI and GPT-5 Thinking: When OpenAI Pauses to Reason

The big idea

GPT-5 Thinking pauses to reason step by step before answering. It's slower but way better at math, code, and tricky logic. Use it when the answer matters more than the speed.

Some examples

Use Thinking for AP Calc problems — it shows its work.
Use standard GPT-5 for chat, drafting, and quick answers.
Thinking can take 30+ seconds — chill and let it cook.
Cost more per token — don't burn it on 'what's the weather'.

Try it!

Pick a tough math problem. Run it on standard GPT-5 and Thinking. Compare answers AND the work shown.

GPT-5 Mini vs Flagship: When Smaller Wins

The big idea

most prompts do not need the flagship; benchmark on your real data

Some examples

Use mini for chat support classifying
Reserve flagship for novel research help
Run both on the same 50 examples to decide

Try it!

Open your favorite AI tool and try one of the examples above. Pick the one that matches what you are actually working on this week. Spend 10 minutes, no more. Notice what worked and what did not — that's the real lesson.

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-builders-model-families-AI-and-gpt5-teen

What is a major improvement in GPT-5 compared to GPT-4?
1. It now generates images instead of text
2. It works without an internet connection
3. It can process longer documents with more detail
4. It uses less computational power
For which type of task is GPT-5 specifically recommended in the material?
1. Simple fact-checking
2. Translating single words
3. Writing short social media posts
4. Hard reasoning tasks
Why is it important to know which GPT model version you are using?
1. Some models require payment
2. Only the newest version can be used legally
3. Different versions produce different answers to the same question
4. Model names are randomly assigned
Which company developed GPT-5?
1. Meta
2. Google
3. OpenAI
4. Microsoft
What does the lesson say you should check when using an AI tool?
1. The developer's email address
2. The weather at your location
3. Which model the tool is actually using
4. The tool's color scheme
You are debugging complex code and want the best possible assistance. Based on what you learned, which model should you choose and why?
1. GPT-3, because it is simpler to use
2. GPT-4, because it has been available longer
3. Either model, because they produce identical code
4. GPT-5, because it has better code capabilities than GPT-4
A classmate tells you that GPT-4 and GPT-5 will give exactly the same answer to any question. Based on the lesson, how would you respond?
1. You ask them to prove it with a video game score
2. You agree, because both are made by the same company
3. You ignore the statement entirely
4. You disagree, because the lesson explicitly states that answers differ between models
If you notice that an AI tool's explanations have become more detailed and logically structured, what might this indicate about the tool?
1. Your internet connection has improved
2. It may have been upgraded to a newer model like GPT-5
3. The AI is malfunctioning
4. It has been downgraded to an older model
Two students use the same prompt on different AI tools. One gets a concise answer, the other gets a detailed explanation. What is the most likely reason for the difference?
1. One computer is newer than the other
2. One student typed faster than the other
3. The tools are using different model versions
4. The students are in different rooms
You are working with a very long research paper that needs to be summarized. Which model would the lesson suggest is better suited for this task?
1. GPT-4
2. GPT-5
3. GPT-3.5
4. All GPT models handle long documents equally
Based on the lesson, what is the relationship between GPT-5 and GPT-4?
1. GPT-4 is newer than GPT-5
2. They are made by different companies
3. They are completely unrelated products
4. GPT-5 is an improved version of GPT-4
A developer wants to build an application that solves complex logical problems. Why might checking the model version be critical for their project?
1. Older models are always free
2. Only GPT-3 can solve logical problems
3. Different model versions produce different quality reasoning outputs
4. Model versions do not affect problem-solving ability
What does it mean that GPT-5 is OpenAI's 'flagship' model?
1. It is only available to paying customers
2. It was released first
3. It is the most advanced or premier model they offer
4. It is the smallest model available
When the lesson mentions that you should 'note where the answers differ,' what is the purpose of this exercise?
1. To prove one model is always wrong
2. To understand how model improvements affect outputs
3. To confuse users
4. To waste time
Based on the key takeaways from this lesson, what is the first step a user should take when starting a new AI task?
1. Check which model version the tool is using
2. Close the tool and restart the computer
3. Immediately ask their question without preparation
4. Write the longest possible prompt

← Back to interactive lesson

Tendril · Builders · Model Families

AI model families: GPT-5 and what's new

Understand what makes GPT-5 different from GPT-4 and earlier OpenAI models.

40 min · Reviewed 2026

The big idea

GPT-5 is OpenAI's flagship model. It's smarter at reasoning, handles longer documents, and is better at code. Knowing what version you're using matters because answers differ.

Some examples

Use GPT-5 for hard reasoning tasks
Notice GPT-5's better code than GPT-4
Compare GPT-5 vs GPT-4 on the same prompt
Check which model your tool actually uses

Try it!

Find a hard math or logic question. Ask it to GPT-4 and GPT-5 (if you have access). Note where the answers differ. Even if both right, the explanations vary.

AI and GPT-5: OpenAI's Latest Big Brain

The big idea

GPT-5 is OpenAI's frontier model. It can read images, write code, do math, and reason through hard problems. ChatGPT Plus uses GPT-5 by default for paying users.

Some examples

Upload a photo and ask GPT-5 to identify what's in it.
Use GPT-5 for coding when Sonnet gets stuck.
GPT-5 is great at structured outputs (JSON, tables).
It's slower than smaller GPTs but more accurate on hard stuff.

Try it!

Try the same hard question on GPT-5 and a smaller model (Haiku or 4o-mini). Compare which one nails it.

AI and GPT-5 Thinking: When OpenAI Pauses to Reason

The big idea

GPT-5 Thinking pauses to reason step by step before answering. It's slower but way better at math, code, and tricky logic. Use it when the answer matters more than the speed.

Some examples

Use Thinking for AP Calc problems — it shows its work.
Use standard GPT-5 for chat, drafting, and quick answers.
Thinking can take 30+ seconds — chill and let it cook.
Cost more per token — don't burn it on 'what's the weather'.

Try it!

Pick a tough math problem. Run it on standard GPT-5 and Thinking. Compare answers AND the work shown.

GPT-5 Mini vs Flagship: When Smaller Wins

The big idea

most prompts do not need the flagship; benchmark on your real data

Some examples

Use mini for chat support classifying
Reserve flagship for novel research help
Run both on the same 50 examples to decide

Try it!

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-builders-model-families-AI-and-gpt5-teen

What is a major improvement in GPT-5 compared to GPT-4?
1. It now generates images instead of text
2. It works without an internet connection
3. It can process longer documents with more detail
4. It uses less computational power
For which type of task is GPT-5 specifically recommended in the material?
1. Simple fact-checking
2. Translating single words
3. Writing short social media posts
4. Hard reasoning tasks
Why is it important to know which GPT model version you are using?
1. Some models require payment
2. Only the newest version can be used legally
3. Different versions produce different answers to the same question
4. Model names are randomly assigned
Which company developed GPT-5?
1. Meta
2. Google
3. OpenAI
4. Microsoft
What does the lesson say you should check when using an AI tool?
1. The developer's email address
2. The weather at your location
3. Which model the tool is actually using
4. The tool's color scheme
You are debugging complex code and want the best possible assistance. Based on what you learned, which model should you choose and why?
1. GPT-3, because it is simpler to use
2. GPT-4, because it has been available longer
3. Either model, because they produce identical code
4. GPT-5, because it has better code capabilities than GPT-4
A classmate tells you that GPT-4 and GPT-5 will give exactly the same answer to any question. Based on the lesson, how would you respond?
1. You ask them to prove it with a video game score
2. You agree, because both are made by the same company
3. You ignore the statement entirely
4. You disagree, because the lesson explicitly states that answers differ between models
If you notice that an AI tool's explanations have become more detailed and logically structured, what might this indicate about the tool?
1. Your internet connection has improved
2. It may have been upgraded to a newer model like GPT-5
3. The AI is malfunctioning
4. It has been downgraded to an older model
Two students use the same prompt on different AI tools. One gets a concise answer, the other gets a detailed explanation. What is the most likely reason for the difference?
1. One computer is newer than the other
2. One student typed faster than the other
3. The tools are using different model versions
4. The students are in different rooms
You are working with a very long research paper that needs to be summarized. Which model would the lesson suggest is better suited for this task?
1. GPT-4
2. GPT-5
3. GPT-3.5
4. All GPT models handle long documents equally
Based on the lesson, what is the relationship between GPT-5 and GPT-4?
1. GPT-4 is newer than GPT-5
2. They are made by different companies
3. They are completely unrelated products
4. GPT-5 is an improved version of GPT-4
A developer wants to build an application that solves complex logical problems. Why might checking the model version be critical for their project?
1. Older models are always free
2. Only GPT-3 can solve logical problems
3. Different model versions produce different quality reasoning outputs
4. Model versions do not affect problem-solving ability
What does it mean that GPT-5 is OpenAI's 'flagship' model?
1. It is only available to paying customers
2. It was released first
3. It is the most advanced or premier model they offer
4. It is the smallest model available
When the lesson mentions that you should 'note where the answers differ,' what is the purpose of this exercise?
1. To prove one model is always wrong
2. To understand how model improvements affect outputs
3. To confuse users
4. To waste time
Based on the key takeaways from this lesson, what is the first step a user should take when starting a new AI task?
1. Check which model version the tool is using
2. Close the tool and restart the computer
3. Immediately ask their question without preparation
4. Write the longest possible prompt

← Back to interactive lesson