GPT-5.4 Leaked with 2M Context Next Week
💡GPT-5.4 rumor: 2M tokens + persistent memory next week—redefines long AI tasks!
⚡ 30-Second TL;DR
What Changed
Leaked via engineer's commit in Codex GitHub repo
Why It Matters
Enables handling of massive documents and long interactions without truncation. Persistent memory boosts reliability for AI agents in production workflows.
What To Do Next
Test long-context prompts on GPT-4o to benchmark readiness for GPT-5.4's 2M window.
Key Points
- •Leaked via engineer's commit in Codex GitHub repo
- •Rumored 2 million token context window
- •Introduces persistent state for better memory
- •Expected launch next week
🧠 Deep Insight
Background and context from public sources — not the original article. 8 sources cited.
🔑 Enhanced Key Takeaways
- •GPT-5.4 was accidentally revealed twice through public Codex GitHub pull requests with force push history that is auditable, not merely speculative[5]
- •OpenAI shipped GPT-5.3-Codex on February 5, 2026, just three weeks before the GPT-5.4 leak, representing an accelerated release cadence of five GPT-5.x model versions in seven months[3][5]
- •The leaked references include technical features such as full-resolution vision support (original-resolution image processing without lossy compression) and a fast mode toggle with priority service tiers for tiered inference pricing[5]
- •Prediction markets on Manifold assign GPT-5.4 a 55% probability of release before April 2026 and 74% before June 2026, though some traders note this would require unprecedented acceleration[5]
- •An OpenAI employee accidentally posted and then deleted a screenshot showing GPT-5.4 in the Codex app's model selector, providing direct visual evidence of the model's existence in internal systems[5]
🛠️ Technical Deep Dive
- Full-resolution vision support: Enables processing of detailed diagrams, high-DPI screenshots, and architectural drawings without lossy compression destroying detail[5]
- Fast mode toggle with priority service tiers: Suggests OpenAI is building tiered inference architecture allowing users to pay more for lower latency, aligning with Codex subscription pricing structure[5]
- Persistent state capability: Rumored feature to maintain context across sessions, addressing the model forgetting problem[7]
- 2 million token context window: Reported capability representing significant expansion from previous context limits[7]
🔮 Future ImplicationsAI analysis grounded in cited sources
⏳ Timeline
📎 Sources (8)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
- geeky-gadgets.com — Openai Gpt 54 Leak
- piunikaweb.com — Openai Gpt 5 4 Accidentally Leaked Codex
- zamin.uz — 192777 Openai Tasodifan Yangi Gpt 54 Modelini Korsatib Qoydi
- help.openai.com — 6825453 Chatgpt Release Notes
- awesomeagents.ai — Gpt 5 4 Codex Repo Leak
- manifold.markets — Gpt 54 Openai Release Date
- longbridge.com — 277450972
- theneurondaily.com — President Trump vs Anthropic vs Openai
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: cnBeta (Full RSS) ↗
This is a summary, not the original. Read the source, or get the weekly briefing.
Weekly AI briefing
One email a week. Unsubscribe anytime.