Day-One Opus 5.5, and a Fresh Gemini Lineup

A customer looked at our model picker this week and asked the right question: "the latest Anthropic model here is Opus 4.7 — what about Opus 5.5?" Fair catch. Model releases move fast, and a reviewer tool should keep up. So we did a full sweep — every model, every price, verified against the providers' own catalogs. Here's what changed.

Claude Opus 5.5 and Opus 5, available now

Anthropic's newest flagship, Claude Opus 5.5, is in the picker as of today — we verified it's live on Anthropic's API before listing it. Notably, it's cheaper than its predecessor: $4 per million input tokens and $20 per million output, versus Opus 5's $5/$25. A newer, smarter model at 20% less is the right direction for review workloads, where thoroughness on a high-stakes PR is exactly what you're paying for. Opus 5 joins the list too.

While we were in there we found and fixed an embarrassing bug: our Opus 4.7 card displayed $15/$75. The real price is $5/$25 — it was always billed correctly (your key, your provider's real rates), but the display was wrong, and if that scared you off Opus reviews, our apologies.

Claude Sonnet 5 stays the default — at $2/$10 it's still the best quality-per-dollar reviewer we know of.

The Gemini 3 generation

Our Gemini list was two generations behind; it now carries the current family:

• Gemini 3.1 Pro (preview) — Google's strongest, $2/$12 (rises to $4/$18 on prompts over 200K tokens).
• Gemini 3.8 Flash — the new default. Google pitches it as "engineered for long-horizon software engineering," which is precisely a code reviewer's job, and it's on intro pricing ($0.75/$3.75) through December 2026.
• Gemini 3.5 Flash-Lite — the budget pick at $0.30/$2.50.

If you had a Gemini 2.5 model selected, nothing breaks — your stored choice keeps working until you switch.

Grok was already fresh

Native xAI support shipped this week with the current lineup (Grok 4.7, Grok 4.3 with its 1M-token context, Grok Build 0.1), so no changes there — and the models page now shows all five providers side by side.

Every price on our model cards is verified against the provider's published rates, and what you actually pay is always your provider's real bill — CodeMouse records the true per-review cost and adds no markup.

FAQ

Do I need to do anything to get the new models?

No. Open Dashboard → API Keys and the new options are in each provider's model picker. Your current selection is unchanged until you switch.

Is Opus 5.5 really cheaper than Opus 5?

Yes — $4/$20 versus $5/$25 per million tokens, per Anthropic's published API pricing. Newer and cheaper.

What happens to my selected Gemini 2.5 model?

It keeps working. The picker now offers the Gemini 3 generation, but stored selections are honored until you change them.

How do you keep prices accurate now?

Model cards are verified against provider catalogs on every lineup change, and your dashboard always records the actual cost of each review from real token usage — the card is a hint, the recorded cost is the truth.