# 🚀 Claude Opus 4.7 is now live on Pickaxe **URL:** \ **Category:** Releases & Features\ **Tags:** claude\ **Created:** [April 16, 2026, 5:44pm UTC](https://community.pickaxe.co/t/claude-opus-4-7-is-now-live-on-pickaxe/8441 "2026-04-16T17:44:32Z")\ **Posts on this page:** 5\ **Page:** 1
**Author:** ![abhi](https://yyz2.discourse-cdn.com/flex004/user_avatar/community.pickaxe.co/abhi/32/5009_2.png) [@abhi](https://community.pickaxe.co/u/abhi)\ **Post date:** [April 16, 2026, 5:44pm UTC](https://community.pickaxe.co/t/claude-opus-4-7-is-now-live-on-pickaxe/8441/1 "2026-04-16T17:44:32Z")
We’ve just added **Claude Opus 4.7** to the model lineup for all Pickaxe builders. ![CleanShot 2026-04-16 at 13.43.51@2x](https://canada1.discourse-cdn.com/flex004/uploads/pickaxeproject/original/2X/e/ee6031498e2aa2394fa75f670055516176da7a45.jpeg) If you’re already building tools, this one is worth trying. ### What’s new (quickly) Claude Opus 4.7 pushes things forward in a few meaningful ways: - Stronger reasoning on complex, multi-step tasks - Better at coding, structured outputs, and long workflows - Improved reliability over longer conversations - More consistent tone and instruction-following It’s the kind of upgrade you _feel_ when your tool just “gets it right” more often. ### Why this matters for Pickaxe builders Pickaxe is all about turning prompts into real tools. With Opus 4.7, a few things get noticeably better: - **More accurate tools** → fewer edge-case failures - **Better Actions + workflows** → especially if you’re generating PDFs, structured data, or chaining tasks - **Stronger knowledge-based responses** → works really well with your Knowledge Base - **Cleaner outputs for end users** → less prompt tweaking needed If you’ve built something that _almost_ worked perfectly before… this model might get you there. ### Where to try it Just head to your **Pickaxe Edit Agent → Prompt** and then scroll down to **AI Model dropdown** and select **Claude Opus 4.7**. ![Claude 4.7 opus](https://canada1.discourse-cdn.com/flex004/uploads/pickaxeproject/original/2X/5/5ba0a1824d854fbdf652544f29d8fe45f08afd2f.gif) Would love to hear what you build with it. If you test it on an existing Pickaxe and notice improvements (or issues), drop it here 👇 ---
**Author:** ![brandoo](https://avatars.discourse-cdn.com/v4/letter/b/7ba0ec/32.png) [@brandoo](https://community.pickaxe.co/u/brandoo)\ **Post date:** [April 17, 2026, 4:55pm UTC](https://community.pickaxe.co/t/claude-opus-4-7-is-now-live-on-pickaxe/8441/2 "2026-04-17T16:55:21Z")
A problem with the pricing ? It was announced like one of the cheapest ---
**Author:** ![carsondev](https://yyz2.discourse-cdn.com/flex004/user_avatar/community.pickaxe.co/carsondev/32/4588_2.png) [@carsondev](https://community.pickaxe.co/u/carsondev)\ **Post date:** [April 17, 2026, 7:05pm UTC](https://community.pickaxe.co/t/claude-opus-4-7-is-now-live-on-pickaxe/8441/3 "2026-04-17T19:05:41Z")
hi ! @brandoo when we announce new models, sometimes it takes a few hours for pricing data to populate in the builder: but it should be showing now ! ![image](https://canada1.discourse-cdn.com/flex004/uploads/pickaxeproject/original/2X/0/06d97171e4439ca45cf645e1c7bb8c80a99c7804.png) ---
**Author:** ![brandoo](https://avatars.discourse-cdn.com/v4/letter/b/7ba0ec/32.png) [@brandoo](https://community.pickaxe.co/u/brandoo)\ **Post date:** [April 17, 2026, 11:55pm UTC](https://community.pickaxe.co/t/claude-opus-4-7-is-now-live-on-pickaxe/8441/4 "2026-04-17T23:55:06Z")
hello @carsondev I meant why Opus 4.7 is at $0.02 per 100uses. It’s a discount at the release of new model from Anthropic or an UI bug ? ---
**Author:** ![lindsay\_support](https://yyz2.discourse-cdn.com/flex004/user_avatar/community.pickaxe.co/lindsay_support/32/4508_2.png) [@lindsay\_support](https://community.pickaxe.co/u/lindsay_support)\ **Post date:** [April 20, 2026, 8:07pm UTC](https://community.pickaxe.co/t/claude-opus-4-7-is-now-live-on-pickaxe/8441/6 "2026-04-20T20:07:02Z")
Hi @brandoo! The estimate you see there is created by taking the average of the cost of each prompt run using that model from across the platform. As you can imagine, the cost of a prompt run can vary a lot depending on the number of tokens used - so initially, right after a model is added, the averages might look a bit off because these averages are generated using just a couple of prompt runs (in this case, runs that used very few tokens). It’s good to give it a few days so users across the platform have a chance to use the new model and get us a good sample of prompt runs 🙂