ChatGPT review
A general-purpose AI assistant for writing, research, coding, analysis, and everyday work.
TECHi verdict
ChatGPT is still the fastest-adopted software product in recorded history — 900M weekly users, a 1B-monthly-active mobile app, and 92% of the Fortune 500 as customers. It is also still a probabilistic text generator that states falsehoods confidently, especially on niche and recent topics. Plus at $20 remains the genuine sweet spot for daily knowledge work; the $8 Go tier (which may carry ads) lowers the paid on-ramp, and the $100 Pro tier finally fills the canyon below the $200 flagship. The honest competitive read: Claude wins on writing and agentic coding, Gemini on price and bundling, Perplexity on cited research. ChatGPT no longer leads any single capability decisively — but nothing else matches its breadth, and breadth is what a default needs.
Pros
- + Most complete all-around assistant, with the deepest tool ecosystem: Search, Deep Research, Agent, Codex, Voice, Memory, Canvas
- + 900M weekly users and the first app to 1B monthly actives — being the default has network effects rivals can't buy
- + Free tier is genuinely useful, and the $8 Go tier is the cheapest paid OpenAI on-ramp yet
- + The GPT-5.6 family (Sol, Sol Pro, Terra, Luna) spans fast chat to Pro-grade reasoning inside one product
- + Connectors cover the main work surfaces — Google Drive, Microsoft 365, Slack, GitHub, Linear, Figma — with OAuth-scoped access
- + Business and Enterprise exclude customer data from model training by default, backed contractually
Watchouts
- - Hallucinations persist across every model tier — confident-sounding false statements, worst on niche and recent topics
- - Sycophancy: the model tends to validate users and incorrect assumptions rather than push back, masking unreliability
- - Pro at $100–$200/month is hard to justify outside heavy research and Codex-intensive workflows
- - Long-context attention degrades — the "lost in the middle" problem still hits very long inputs
- - Active copyright litigation (NYT lawsuit, German lyrics ruling) leaves training-data legality unsettled
- - Training-data use is opt-out on individual tiers, and Go may include ads — the data tradeoffs live in the fine print
Frequently asked
What are the pros of ChatGPT?
Most complete all-around assistant, with the deepest tool ecosystem: Search, Deep Research, Agent, Codex, Voice, Memory, Canvas. 900M weekly users and the first app to 1B monthly actives — being the default has network effects rivals can't buy. Free tier is genuinely useful, and the $8 Go tier is the cheapest paid OpenAI on-ramp yet. The GPT-5.6 family (Sol, Sol Pro, Terra, Luna) spans fast chat to Pro-grade reasoning inside one product. Connectors cover the main work surfaces — Google Drive, Microsoft 365, Slack, GitHub, Linear, Figma — with OAuth-scoped access. Business and Enterprise exclude customer data from model training by default, backed contractually
What are the watchouts / cons of ChatGPT?
Hallucinations persist across every model tier — confident-sounding false statements, worst on niche and recent topics. Sycophancy: the model tends to validate users and incorrect assumptions rather than push back, masking unreliability. Pro at $100–$200/month is hard to justify outside heavy research and Codex-intensive workflows. Long-context attention degrades — the "lost in the middle" problem still hits very long inputs. Active copyright litigation (NYT lawsuit, German lyrics ruling) leaves training-data legality unsettled. Training-data use is opt-out on individual tiers, and Go may include ads — the data tradeoffs live in the fine print
What is the TECHi verdict on ChatGPT?
ChatGPT is still the fastest-adopted software product in recorded history — 900M weekly users, a 1B-monthly-active mobile app, and 92% of the Fortune 500 as customers. It is also still a probabilistic text generator that states falsehoods confidently, especially on niche and recent topics. Plus at $20 remains the genuine sweet spot for daily knowledge work; the $8 Go tier (which may carry ads) lowers the paid on-ramp, and the $100 Pro tier finally fills the canyon below the $200 flagship. The honest competitive read: Claude wins on writing and agentic coding, Gemini on price and bundling, Perplexity on cited research. ChatGPT no longer leads any single capability decisively — but nothing else matches its breadth, and breadth is what a default needs.