OpenAI shipped GPT-5.5 on Apr. 23, pitching the model, codenamed "Spud," as its sharpest system yet for autonomous, multi-step work.
GPT-5.5 Agentic Coding Push
The release lands exactly one week after Anthropic pushed Claude Opus 4.7 into general availability, setting up a direct head-to-head on agent workloads, per TechCrunch and Fortune.
GPT-5.5 is built to plan, run tools, check its own output, and iterate without constant prompting.
President Greg Brockman called it "a new class of intelligence" on a call with reporters, framing it as a step toward "more agentic and intuitive computing."
The model rolls out to ChatGPT Plus, Pro, Business, and Enterprise, with a heftier Pro variant also live. API pricing starts at $5 per million input tokens and $30 per million output tokens over a one-million-token context window.
Also Read: Ethereum Nears $2,450 Showdown As Bulls And Bears Split On Next Move
Opus 4.7 Benchmark Gap
OpenAI's own numbers, which VentureBeat flagged as state of the art on 14 evals, place GPT-5.5 at 82.7% on Terminal-Bench 2.0, well clear of Opus 4.7's 69.4%.
On FrontierMath Tiers 1 through 3, the new model hit 51.7%, versus 43.8% for Anthropic's flagship.
Computer-use scores ran tighter, with GPT-5.5 at 78.7% on OSWorld-Verified against Opus 4.7 at 78.0%, though GPT-5.5 Pro pulled clear on browsing at 90.1% to 79.3%.
Reviewers still credit Opus 4.7 with stronger research writing and tighter instruction-following, plus higher-resolution vision at roughly 3.75 megapixels.
The cadence keeps tightening. GPT-5.5 arrives six weeks after GPT-5.4. Anthropic pushed Opus 4.6 in February before Opus 4.7, while Google has kept Gemini 3.1 Pro in the same enterprise bracket.
Read Next: TRON Connects $85B USDT Network To LI.FI In Cross-Chain DeFi Push