ChatGPT vs Gemini

A detailed comparison of AI assistant subscriptions and performance

Feature ChatGPT Gemini
Free Go Plus Free Plus Pro
Reasoning model limit per day 1[GPT 5 Thinking Mini]¹ 10[GPT 5 Thinking Mini]¹ 428(3000/week)[GPT 5.2 Thinking] 3+9[Gemini 3 Pro + Flash Thinking] 30+90[Gemini 3 Pro + Flash Thinking] 100+300[Gemini 3 Pro + Flash Thinking]
Non-reasoning model limit 40(10 every 5h)[GPT 5.2 Instant] Unlimited[GPT 5.2 Instant] Unlimited[Gemini 3 Flash]
Context window limit 16k / 196k(Non-reasoning / Reasoning) 32k / 196k(Non-reasoning / Reasoning) 32k 128k 1M
File uploads per day 3 30 640(80 every 3h) Unknown
Memory features Memories Memories Memories, history, search Memories, chat history
Image generation 5?²[GPT Image 1.5] >5?²[GPT Image 1.5] 50?²[GPT Image 1.5] 3+100[Nanobanana Pro + Nanobanana] 30+300?[Nanobanana Pro + Nanobanana] 100+1000[Nanobanana Pro + Nanobanana]
Image resolution (1:1) 1024×1024 (1K)[GPT Image 1.5] 2048×2048 (2K) / 1024×1024 (1K)[Nanobanana Pro / Nanobanana]
Custom model creation No[Custom GPTs] Yes[Custom GPTs] Yes[Gems]
Folder organization Yes[Projects] No
Deep research per month 5[GPT 5.2 Thinking] 5³[GPT 5.2 Thinking] 25[GPT 5.2 Thinking] 5[Gemini 3 Flash Thinking] 90(3/day)[Gemini 3 Pro] 600(20/day)[Gemini 3 Pro]
Agent mode per month No No 40 No
Advertisements Yes Yes No No
Video processing per day No 3(Up to 5 minutes per file) 6(Up to 5 minutes per file) 20(Up to 60 minutes per file)
Audio processing per day No 3(Up to 10 minutes per file) 6(Up to 10 minutes per file) 20(Up to 180 minutes per file)
Youtube summaries No Yes
Cloud storage included No 15GB 200GB 2TB
Plan sharing No No Up to 6 accounts
Best option
Sufficient (subjectively)

Footnotes

  • Free and Go accounts don't have access to GPT 5.2 Thinking, so it's likely they use GPT 5 Thinking Mini instead.
  • GPT Image 1.5 limits are unknown. GPT Image 1 limits were ~50/day for Plus, ~5/day for Free.
  • According to user reports. The actual limits are not public.
  • Open AI has only reported that deep research uses a thinking model, it is safe to assume it's the latest one.
  • Ads only in the USA, for now.
  • While Gemini has a memory feature, it sometimes calls memory when not needed, or decides not to call it when actually needed; which sometimes provokes inconsistent outputs.
  • Estimated based on Plus reasoning model limits being ~1/3 of Pro limits.
  • NotebookLM differs from Projects. While it can be linked with Gemini chats, it lacks a folder structure when used insideGemini; and when used on the NotebookLM site it is less creative due to an anti-hallucination tuning.
  • Free accounts use the Google account's free 15GB storage.

General Accuracy — AA-Omniscience Benchmark

Claude models

Higher is better for Correct & Abstention. Lower is better for Hallucination.

AA-Omniscience Notes

  • Correct answers were calculated with Correct = A.
  • Abstention answers were calculated with Abstention = (1−A)⋅(1−H).
  • Hallucinations answers were calculated with Hallucination = (1−A)⋅H.
  • (Where "A" is the percentage scored on the "AA-Omnisicence Accuracy" benchmark, "H" is the percentage scored on the "AA-Omnisicence Hallucination Rate" benchmark.)

Long Context Accuracy — MRCR v2 8-Needle Benchmark

Claude models

Shows how accuracy degrades as context length increases.

MRCR v2 8-Needle Notes

  • • There's no benchmark for 196k tokens, the next available one is 256k, which is not supported by ChatGPT on the web and app.
  • • Maximum context window of GPT 5.2 Thinking is 196k tokens.
  • • Maximum context window of GPT 5.2 Instant is 16k/32k tokens (Free/Plus & Go).
  • • Maximum context window of Gemini 3 Pro, Gemini 3 Flash Thinking and Gemini 3 Flash is 32k/128k/1M tokens (Free/Plus/Pro).
  • • Maximum context window of Claude Opus 4.5 and Claude Sonnet 4.5 is 200k tokens.

Notes

  • • For "GPT 5.2 Thinking", "gpt-5.2:medium" was used in the graph; as GPT-5.2:xhigh is not available on the web.
  • • For "GPT 5.2 Instant", "gpt-5.2" was used in the graph.
  • • For "Gemini 3 Pro", "gemini-3-pro-preview:high" was used in the graph.
  • • For "Gemini 3 Flash Thinking", "gemini-3-flash-preview:thinking:high" was used in the graph.
  • • For "Gemini 3 Flash", "gemini-3-flash-preview" was used in the graph.
  • • For "Claude Opus 4.5 Thinking", "claude-opus-4.5:thinking" was used in the graph.
  • • For "Claude Sonnet 4.5 Thinking", "claude-sonnet-4.5:thinking" was used in the graph.
  • • Claude models included due to them being available for Gemini users through the Antigravity app.

ChatGPT

  • Web search thinks significantly longer (can easily take 7-10 min vs Gemini's 1-3 min)
  • Easy switching between different personalities
  • Can connect to a limited selection of sites, though tool calls can be inconsistent
  • Offers integration with Github, through the Codex site and the Visual Studio Code extension

Gemini

  • Deep research produces longer, more detailed investigations
  • Nanobanana Pro offers faster image generation than GPT Image 1.5. Its images also appear more realistic (subjectively)
  • Native Android integration (iOS confirmed to come soon), though it can be over-eager with tool calls
  • Offers uses of Claude Opus 4.5 and Claude Sonnet 4.5 through the Antigravity app