Opus 4.6 & Sonnet 4.6: Two New Models, Bigger Projects

Primio now runs on both Claude Opus 4.6 and Claude Sonnet 4.6 — Anthropic’s two newest models. We also rebuilt the entire conversation engine to handle bigger projects, longer sessions, and more complex apps.

Build Bigger Projects

The old limits are gone. We’ve completely rearchitected how Primio handles your conversations:

  • Longer sessions — Work on complex apps for hours without losing context
  • Automatic memory optimization — Primio now summarizes older parts of your conversation automatically, keeping what matters while freeing up space for new decisions
  • More stable on big projects — Apps with dozens of screens now build reliably from start to finish

No more hitting walls halfway through a complex build.

Opus 4.6: Smarter Model, Better Code

Opus 4.6 brings real improvements to the code Primio writes:

  • Thinks before it codes — Plans architecture decisions before writing, leading to cleaner results
  • Catches its own mistakes — Errors get fixed before they reach your preview
  • Stays focused — Multi-step features work more reliably the first time
  • 91.3% on GPQA Diamond — a major jump in deep reasoning over the previous generation

You’ll notice it in the details: better widget choices, smarter state management, fewer “that’s not what I meant” moments.

Sonnet 4.6: Near-Opus Performance at a Fraction of the Cost

Sonnet 4.6 closes the gap with Opus almost entirely on coding tasks — and it’s noticeably faster:

  • 79.6% on SWE-bench Verified — within 1.2% of Opus 4.6’s score
  • 72.5% on OSWorld — near-identical computer use performance to Opus 4.6 (72.7%)
  • 4x abstract reasoning improvement — ARC-AGI-2 jumps from 13.6% to 58.3% vs the previous generation
  • Faster generation — snappier responses on every turn

For most projects, Sonnet 4.6 delivers Opus-quality results at significantly lower token cost. Use it as your everyday workhorse.

Faster Conversations

Open any project and your conversation loads immediately — even projects with hundreds of messages. No more waiting for long threads to populate.

What This Means

  • Bigger apps — Build complex, multi-screen apps without hitting limits
  • Longer sessions — Iterate for hours without context degradation
  • More stability — Fewer interruptions on ambitious projects
  • Better code — Cleaner architecture, fewer bugs
  • Model choice — Pick Opus for maximum reasoning, Sonnet for speed and efficiency

Try It Now

Both models are live. Use the model selector in any project to switch between Auto, Opus 4.6, Sonnet 4.6, and Haiku 4.5. Auto picks the best model for each task automatically.

Start a new project or continue an existing one — the improvements are immediate.

Build better native mobile apps. Build them with Primio.