Claude Code just got cheaper, faster and a lot more customisable
Claude Code just got cheaper, faster and a lot more customisable
Zoby OS | 3 October 2026
Anthropic's latest Claude Code update dropped this week, and there's more in it than usual. Two new models, a brand-new way to customise the tool, and a couple of freebies with deadlines attached. Here's what matters and what to do about it.
The 20-second version
Opus 5.5 is the new default model. Roughly as capable as Fable 5.1, over 30% faster and 40% cheaper to run.
Sonnet 5.5 is out too. The quicker, lighter option for clear, well-defined jobs.
"Mods" let you change how Claude Code itself works, and Claude can write them for you.
Free stuff: bonus credit for cloud sessions (claim by 7 October) and a one-off usage reset (use by 22 October).
The new models
Opus 5.5 is the headline. It performs at roughly the level of Fable 5.1 on most work, but it's faster and noticeably cheaper. Anthropic also says it gives shorter replies, sticks more closely to what you actually asked, and uses far fewer "Claudisms", including those long dashes that make text scream "an AI wrote this". Good news for anyone tired of editing them out.
Sonnet 5.5 is the workhorse. Over 30% faster than Sonnet 5 and uses fewer tokens (the units AI usage is measured in) per task. Best for jobs where you know exactly what you want, like fixing a specific bug.
Our take: "Same brain, lower bill" is the most useful kind of AI upgrade. If cost has been putting you off heavier AI use, this nudges the maths in your favour.
Mods: build the feature you wish existed
A mod is a small add-on that hooks into Claude Code and changes how it behaves. It can add a new command, show extra information on screen, or swap out a built-in feature for your own version.
The clever bit is you don't need to write it. You describe what you want in plain English and Claude builds the mod and switches it on.
Our take: Most event professionals won't touch this directly yet. But it's a sign of where AI tools are heading: less "one size fits all", more "shape it around how you work".
Other changes worth knowing
No more half-finished jobs. If you hit your five-hour usage limit mid-task, Claude now borrows a small amount from your weekly allowance to finish what it started, rather than stopping dead.
Cloud sessions are now part of your plan. These run Claude Code on Anthropic's servers instead of your laptop, so work keeps going when you close the lid.
Five-hour limits are up 20%. More room before you hit the wall.
Anyone can now publish plugins to the Claude directory. These reach everyone using Claude, not just developers, so expect more ready-made tools to appear over the coming months.
A built-in "health check" for your setup. A new command reviews your instructions and flags anything written for older models or that contradicts itself.
What to do this week
→ Claim your bonus credit by 7 October. Existing Pro subscribers get $100, Max subscribers get $250, to spend on cloud sessions. Run /claim-credit in Claude Code.
→ Use your free usage reset by 22 October. Everyone on Pro, Max and Team gets one full reset, sitting in Settings > Usage.
→ Re-check your saved instructions. Two model launches in ten days means older prompts may need a tidy. If you use Claude Code, run /checkup prompt-audit.
→ If cost held you back, try again. Opus 5.5 is a good excuse to revisit tasks you shelved as too expensive.
The bottom line: this update is about making powerful AI cheaper, quicker and more personal. For events teams, the immediate win is the free credit and reset, so grab them before they expire.