ChatGPT-5.5 Outperforms Claude and Gemini

Welcome back to The Shift AI. Let’s get straight to what matters in AI today…

Today we have:

🤖 OpenAI Just Released GPT-5.5 and It Is Setting New Benchmarks Across the Board

📹How to Turn Long Videos Into Viral Clips With OpenShorts

🤖 DeepSeek Is Back and It Just Released a New Model With Drastically Lower Costs

🔨Tools and Shifts you cannot miss

🤖 OpenAI Just Released GPT-5.5 and It Is Setting New Benchmarks Across the Board

OpenAI released GPT-5.5 this week, its smartest model yet, setting new highs across coding, reasoning, computer use, and knowledge work benchmarks, with several scores comparable to Claude Mythos, at half the cost of competitive frontier coding models.

The Shift:

What Changed From GPT-5.4 - GPT-5.5 matches GPT-5.4's per-token latency while performing at a significantly higher intelligence level. It uses fewer tokens to complete the same Codex tasks, making it more efficient and more capable at the same time.

Better Than Everything Else Right Now - GPT-5.5 outperforms Claude Opus 4.7 ( 82.7% vs Opus 4.7's 69.4%) and Gemini 3.1 Pro across coding, computer use, reasoning, and knowledge work benchmarks. Several scores come close to Claude Mythos, which remains behind a restricted access program. 

What It Actually Does in Practice - OpenAI's finance team used it to review over 24,000 tax forms, finishing two weeks faster than the prior year. A go-to-market employee automated weekly business reports and saved 5 to 10 hours every week.

Pricing and Availability - GPT-5.5 is priced at $5 per million input tokens and $30 per million output tokens, half the cost of competitive frontier coding models. Rolling out now to Plus, Pro, Business, and Enterprise users in ChatGPT and Codex, with API access coming very soon.

Senior engineers who tested GPT-5.5 said it was noticeably stronger than GPT-5.4 and Claude Opus 4.7 at reasoning and autonomy, catching issues in advance without being prompted. Cursor's CEO called it noticeably smarter and more persistent. One NVIDIA engineer said losing access felt like having a limb amputated.

Together with Wispr Flow

Talk to your AI tools the way you'd talk to a colleague.

You don't send a colleague a three-word brief. You explain the context, the constraints, what you've already tried. But typing all that into ChatGPT takes forever — so you don't.

Wispr Flow lets you speak your prompts instead. Talk through your thinking naturally and get clean, paste-ready text. No filler words. No cleanup. Just detailed prompts that actually get you useful answers on the first try.

Millions of users worldwide. Works system-wide on Mac, Windows, and iPhone.

Try Wispr Flow free

📹How to Turn Long Videos Into Viral Clips With OpenShorts

No subscription. No watermarks. No limits. Here’s how it works.

Step 1: Self-host with Docker

Clone from GitHub and run locally. Free forever, no usage caps, no platform fees.

Step 2: Drop in your video

Paste a YouTube URL or upload a local file into the Clip Generator.

Step 3: Gemini finds the best moments

Google Gemini analyzes the transcript and detects 3 to 15 high-potential viral moments automatically. No manual scrubbing.

Step 4: Auto-crop, subtitles, and hooks

Every clip gets reframed to 9:16, word-level subtitles burned in, and AI-generated hook text overlaid, automatically.

Step 5: Publish

Post directly to TikTok, Instagram Reels, and YouTube Shorts in one click.

What clipping tools charge $49/month for, OpenShorts does for free. You can get it on GitHub.

🤖 DeepSeek Is Back and It Just Released a New Model With Drastically Lower Costs

More than a year after the original DeepSeek shock rattled the AI industry, the Chinese startup released DeepSeek-V4 this week, an open source model with ultra-long context and drastically reduced costs that experts are calling an industry inflection point.

The Shift:

Two Versions, One Big Claim - DeepSeek-V4 comes in two versions, V4-Pro for maximum capability and V4-Flash for efficiency and lower cost. On the world knowledge reasoning benchmark, V4-Pro trails only Google's latest Gemini model, a strong claim for an open source release.

Ultra-Long Context Is the Real Story - V4 addresses a long-standing industry problem, slower performance, and higher costs at long context lengths. Experts say this is a genuine inflection point, making long-text processing accessible at scale for the first time without the usual trade-offs.

Built for AI Agents and Chinese Chips - V4 is optimized for popular AI agent tools, including Claude Code, OpenClaw, and CodeBuddy. It also runs on Huawei's Ascend SuperPoD chips, significant given Huawei has been sanctioned by the US since 2019 over national security concerns.

The Geopolitical Backdrop Is Impossible to Ignore - The White House accused Chinese entities of a massive AI technology theft effort the day before this release. Beijing called the claim baseless. DeepSeek's original model was called a Sputnik moment for the industry.

A preview version of the open source model is available now, with no full release date confirmed. Data privacy and censorship concerns from the original DeepSeek launch remain unresolved; the chatbot still refuses to answer questions on sensitive topics.

Together with Unblocked

[Webinar] Stop babysitting your coding agents

MCPs give your agents access to information, not understanding. The teams pulling ahead are using a context engine to give agents the right context for every task, so they stay on track without the set up tax or the correction loops. Join live on May 6 (FREE) to see how.

Register now

🔨AI Tools for the Shift

📊 Alfina AI - Learn investing fundamentals, discover stock ideas, and generate research-driven analysis reports with AI.

✍️ Wordobot - Create highly personalized, persuasive content by analyzing customer needs, fears, and desires with AI.

🔎 Parallel Web Systems - Power your AI with highly accurate web search results designed for reliable, real-time information.

💼 Lighthouse - Automate venture deal flow from sourcing to due diligence with AI-driven insights and workflows.

📈 Flowsery - Turn AI into your analytics assistant to track traffic, revenue, and goals across all your websites.

🚀Quick Shifts

🔗 Claude now connects with personal apps like Spotify, Uber Eats, and TurboTax, enabling in-chat recommendations and actions, while keeping user data private and not using it for model training.

📉 Meta plans to lay off about 10% of its workforce, roughly 8,000 employees, and close 6,000 open roles as it reallocates resources toward heavy AI investments and infrastructure expansion.

📰 Noscroll, a new AI startup, has launched a bot that scans social media, news sites, and forums, delivering curated summaries and breaking alerts via text so users can stay updated without scrolling.

🏥 OpenAI launched ChatGPT for Clinicians, a free tool for verified U.S. healthcare professionals, with GPT-5.4 scoring 59.0 on HealthBench Pro, outperforming physicians and Anthropic’s Opus 4.7.

🚀 SpaceX is reportedly in talks with French AI startup Mistral for a potential three-way partnership, alongside its recent collaboration with coding platform Cursor to expand its AI capabilities.

🧩 Prompt of the Day

How to Run a Joint Email Campaign That Expands Reach

Most collaborations stop at shoutouts. The real leverage comes from combining audiences with aligned messaging and structured campaigns.

Turn partnerships into shared growth engines.

Paste the prompt: Drop this into ChatGPT, then fill in your partner details.

Prompt to paste

Create a joint email campaign strategy for [Insert brand] and [Insert partner brand]. Define messaging themes that align both audiences, outline a campaign schedule, and segment users for targeted sends. Focus on expanding reach, increasing engagement, and driving conversions from both lists.

🤳AI Nugget of the Day

That’s all for today’s edition. See you tomorrow as we track down and get you all that matters in the daily AI Shift!

If you loved this edition, let us know how much:

Forward it to your pal to give them a daily dose of the shift so they can 👇