My Cheatsheets

Scope note: This guide reflects observed behavior and published limits for ChatGPT and major frontier models as of mid-2026. Treat numeric limits as indicative; vendors change them without notice. For current ChatGPT plan limits, see OpenAI's help center.

Core Principles

• LLMs are predictors, not logic engines. They generate plausible continuations, not guaranteed truths.
  • For additional information about how this works, see How LLMs Actually Generate Text (Every Dev Should Know This).
• Think of them as brilliant, distractible assistants. Fast and knowledgeable, but prone to drift, agreement bias, and subtle hallucination.

Tokens & Context

• Context windows continue to expand, but practical reliability still degrades near the ceiling.

  • GPT‑5.5 (ChatGPT): 128k tokens (Instant and Thinking); 272k for GPT-5.5 Pro. GPT-5.2 retired June 12, 2026.
    • GPT-4.1 family: up to ~1M tokens in API contexts (retired from ChatGPT UI Feb 2026).
  • Claude 4 (Opus/Sonnet): ~200k–1M tokens depending on plan and variant.
  • Gemini 2.5 Flash / Pro: ~1M tokens input; output typically capped ~65k (Flash documented).
  • Qwen 2.5 (72B Instruct): ~32k tokens (open model cards).
  • Perplexity Sonar: ~128k tokens context reported.

• Context drops still occur when limits are exceeded; older messages are truncated or summarized.

• Best practice: never rely on “infinite” context. Chunk work, summarize checkpoints, and externalize state into documents.

ChatGPT Usage Limits

[Updated June 2026] GPT-5.2 was retired June 12, 2026. The current model family is GPT-5.5 (Instant, Thinking, Pro). Canvas was removed from GPT-5.5 on May 28, 2026; writing and coding now use inline writing blocks and code blocks. For current per-plan limits, see OpenAI’s help center — numbers change frequently.

Key distinctions that remain stable

• Manual Thinking requests count toward weekly limits.
• Automatic internal escalation from Instant mode does not consume manual Thinking quota.
• UI limits are often lower than API limits.
• Model picker labels (Instant / Thinking / Pro) map to GPT-5.5 variants; Thinking Light was removed.

Model Behavior

What’s Stable

• Focused sessions outperform sprawling threads.
• Explicit structure (steps, constraints, deliverables) improves consistency.
• Repeating key facts or invariants reduces drift.

What’s Changed

• Canvas removed (May 2026): Writing and coding no longer use a side panel. Output now appears as inline writing blocks and code blocks in the chat thread. Version history and side-by-side editing are gone; externalize long-form work to documents.

• Memory features are now broadly enabled:

  • Preferences and style can persist across sessions.
  • Memory can usually be reviewed, edited, or wiped in settings.

• Safety tuning is stronger and more dynamic:

  • More hedging.
  • More refusals around edge cases.
  • Behavior can change week to week without version bumps.

Confidence vs Accuracy

• Models remain agreeable and can reinforce your mistakes.
• Hedging has increased, but correctness has not improved proportionally.

Rule of thumb:

• Trust for brainstorming, outlining, summarization, code scaffolding.
• Verify for math, logic, legal, medical, financial, or historical facts.

Choosing the Right Model (mid-2026 snapshot)

Model specs change rapidly. Treat this as a starting point; verify current context windows and limits with each provider.

---------------- | -------------------------- | ------------- | ------ | ------- | ----------------------------- | | GPT‑3.5 | ~16k | ~4k–8k | No | Limited | Fast, shallow; largely legacy | | GPT‑4‑Turbo | ~128k | ~8k–16k | Yes | Yes | Stable baseline | | GPT‑4.1 family | Up to ~1M (enterprise/API) | ~32k | Yes | Yes | Availability varies by plan | | GPT‑5.2 | ~256k standard | ~32k (varies) | Yes | Yes | Current flagship | | GPT‑4o / 4o‑mini | ~128k | Lower in UI | Yes | Yes | Fast, multimodal | | Claude 4 | ~200k–1M | ~64k | Yes | Varies | Plan‑dependent | | Gemini 2.5 Flash | ~1M input | ~65k | Yes | Yes | Speed‑optimized, API‑centric | | Gemini 2.5 Pro | ~1M (reported) | Unclear | Yes | Yes | Output caps vary | | Qwen 2.5 | ~32k | Unclear | Varies | Varies | Open model cards | | Perplexity Sonar | ~128k | Unclear | Varies | Yes | Search‑centric |

File & Data Handling

• Uploading many files increases confusion and token pressure.

• Best practice:

  • Work with 2–3 files at a time.
  • Use staged comparisons (A vs B, then fold in C).

• Modality quality matters:

  • OCR errors propagate.
  • Transcripts drop nuance.

Math & Logic (Still Weak Spots)

• Pattern math is strong; symbolic manipulation is fragile.
• Multi‑step reasoning benefits from explicit scaffolding.

Better approach:

• Ask for executable code.
• Use external solvers or calculators for critical results.

Practical Tips

• Specify output shape and depth explicitly.
• Use headings, checkpoints, and acceptance criteria.
• Externalize long‑running work to shared documents.
• Reset threads when contradictions accumulate.

Memory & Personalization

• Memory is active by default in many systems.
• Review stored memory periodically.
• Store preferences and durable context only.
• Prune aggressively.

Considerations

• Cost: long contexts are expensive with diminishing returns.
• Privacy: many providers train on chats unless you opt out.
• Safety: guardrails shift without notice.

Bottom Line

• Keep chats focused and state explicit.
• Know your model’s current limits.
• Verify anything that matters.
• Treat AI as an assistant—not a calculator, database, or oracle.