In February, we shipped Opus 4.6 on SuperNinja and called it the biggest intelligence upgrade we'd ever rolled out. Anthropic just made us revise that sentence.
Opus 4.7 is live in SuperNinja's Complex Plus mode today. It's faster, sharper, and better at the thing agents actually have to do for a living — staying on task across long, messy, multi-step work.
What's new in Opus 4.7
This isn't a minor version bump. Opus 4.7 is a focused, well-targeted upgrade to the areas that matter most for real work: coding, multi-step reasoning, and long-running agentic tasks.
#1 in Software Engineering
SWE-bench Verified jumps from 80.8% to 87.6%, putting Opus 4.7 ahead of both GPT-5.4 and Gemini 3.1 Pro. On SWE-bench Pro, the harder multi-language benchmark, Opus 4.7 reaches 64.3%, leapfrogging GPT-5.4 at 57.7% and Gemini at 54.2%.
#1 in Tool Use
Opus 4.7 leads MCP-Atlas at 77.3%, ahead of GPT-5.4 (68.1%) and Gemini 3.1 Pro (73.9%). For agents that call tools across complex, multi-step workflows, this is the number that matters.
#1 in Knowledge Work
On GDPval-AA — which tests real-world tasks in finance, legal, and other professional domains — Opus 4.6 outperforms GPT-5.2 by 144 Elo points. That's a massive gap.
Sharper Vision
Opus 4.7 introduces high-resolution vision support up to 3.75 megapixels, a 3x increase over its predecessor. It can now read and reason about dense charts, detailed UI screenshots, and complex visual documents with precision.
Longer, More Reliable Agentic Work
Multi-step agentic reasoning improved 14% with a third of the tool errors. New task budgets for agentic loops let SuperNinja run longer, more complex workflows without going off-track. Sessions that previously lost context across steps now hold it.
What this means inside SuperNinja
SuperNinja's Complex Plus mode is where you send the hard stuff. The multi-file refactor. The research task with fifteen tool calls. The deep-dive analysis across a hundred pages.
Opus 4.7 makes that mode materially better at all of it.
Code & app building. Better architecture on multi-file projects. Cleaner code. Fewer edge cases missed. Full-stack apps that ship on the first run instead of the third.
Deep Research. The tool-use gains compound here. Better navigation, better source extraction, better synthesis across long research sessions.
Slides & presentations. The Agentic Slide Maker gets the reasoning upgrade and the sharper vision. Better narrative structure. Smarter chart picks. Less back-and-forth.
Document & data work. PDFs, spreadsheets, dashboards. The vision upgrade matters most here — Opus 4.7 can actually read the chart it's analyzing.
Long-running automations. Scheduled tasks that run without drift. This is where the 14% improvement on multi-step reasoning and the one-third reduction in tool errors stack up fastest.
How to use Opus 4.7 on SuperNinja
- Open super.myninja.ai.
- Click the model selector.
- Select Complex → Complex Plus.
- Submit your task.
Complex Plus uses more compute than the standard modes. Use it when the task deserves it — building full apps, long research, complex coding, visual analysis, anything that used to take a person a day.
Availability
Opus 4.7 is live today in Complex Plus mode for all SuperNinja users. No waitlist. No separate signup.
If you're already on SuperNinja, it's in the model selector now.
If you're not, the free tier is enough to try it.
