Honestly, I get the hype cuz [[PRODUCT:Anthropic Claude 3.5 Sonnet]] is basically magic for coding, but I'd be super cautious about holding off on big projects! Based on my experience watching these cycles, we probably wont see [[PRODUCT:Anthropic Claude 5]] until well into 2025. Anthropic usually does these mid-tier "pro" updates first, so rushing into a long-term setup expecting a 2024 launch is risky. Plus, the $20/month for [[PRODUCT:Anthropic Claude Pro]] is still the best value right now, so just stick with that and save ur money until an official roadmap actually drops!!

So, I actually went through a whole nightmare last year when I paused a dev project waiting for the 'next big thing' in LLMs, and tbh it was a total waste of time because the landscape shifted so fast that my original architecture became obsolete anyway. Respectfully, I'd consider another option besides just waiting for a specific release like Claude 5.

Hmm, I've had a different experience with these cycles—Anthropic is definitely unpredictable, and honestly, you might be waiting for a phantom date. Be careful about hitting the brakes now because:

- The jump from 3.5 to 4 or 5 usually involves massive architectural changes that might break your current implementation anyway
- You'll miss out on early-mover advantage while others are building on what's available today
- Rumors usually suggest we'll see more incremental 'Opus' or 'Sonnet' refreshes before a full version jump

I really think you should just build for the API you have now and focus on modularity so you can swap later. Waiting for a release that doesnt even have a leaked internal codename yet is risky. Just my two cents tho!

In my experience, I would suggest not getting too hung up on specific release dates because the safety protocols Anthropic follows are pretty intense. I've been tracking these development cycles for years, and they honestly prioritize *alignment* and safety over just rushing a new version out the door. It's basically their whole brand identity.

So, while everyone is guessing about 2025, I’d say just stick with the current brand offerings for now. You really can't go wrong with their ecosystem if you care about reliability and fewer hallucinations. I've seen too many people stall their projects waiting for the "perfect" model only to realize the current tech was already good enough for their production needs. Plus, their track record for gradual safety-first updates means whatever they drop next will be solid. Honestly, just keep building with what they have available now... it's better to have a working prototype than a bunch of "what ifs" in your head lol. Just my two cents!

So, I actually went through a whole nightmare last year when I paused a dev project waiting for the 'next big thing' in LLMs, and tbh it was a total waste of time because the landscape shifted so fast that my original architecture became obsolete anyway. Respectfully, I'd consider another option besides just waiting for a specific release like Claude 5.

Hmm, I've had a different experience with these cycles—Anthropic is definitely unpredictable, and honestly, you might be waiting for a phantom date. Be careful about hitting the brakes now because:

- The jump from 3.5 to 4 or 5 usually involves massive architectural changes that might break your current implementation anyway
- You'll miss out on early-mover advantage while others are building on what's available today
- Rumors usually suggest we'll see more incremental 'Opus' or 'Sonnet' refreshes before a full version jump

I really think you should just build for the API you have now and focus on modularity so you can swap later. Waiting for a release that doesnt even have a leaked internal codename yet is risky. Just my two cents tho!

Story time: I went through this last year waiting for a major drop and ended up blowing my budget on credits while holding out. Honestly, sticking with Anthropic is usually a solid bet for value, but I found that waiting for the next big version kinda backfired when the subscription costs added up without me finishing my project. Now I just dive in with whatever that brand has available because the "next big thing" always feels just out of reach anyway!

So, I actually went through a whole nightmare last year when I paused a dev project waiting for the 'next big thing' in LLMs, and tbh it was a total waste of time because the landscape shifted so fast that my original architecture became obsolete anyway. Respectfully, I'd consider another option besides just waiting for a specific release like Claude 5.

Hmm, I've had a different experience with these cycles—Anthropic is definitely unpredictable, and honestly, you might be waiting for a phantom date. Be careful about hitting the brakes now because:

- The jump from 3.5 to 4 or 5 usually involves massive architectural changes that might break your current implementation anyway
- You'll miss out on early-mover advantage while others are building on what's available today
- Rumors usually suggest we'll see more incremental 'Opus' or 'Sonnet' refreshes before a full version jump

I really think you should just build for the API you have now and focus on modularity so you can swap later. Waiting for a release that doesnt even have a leaked internal codename yet is risky. Just my two cents tho!

bump

+1

Been thinking about this since I saw the thread earlier, and honestly, rmzvjzkkzy is spot on about not waiting around. I used to be a brand loyalist but lately, I've found that running a hybrid setup is way more efficient than waiting for Anthropic Claude 5. If you're doing heavy dev work, I'd suggest checking out these two instead of holding your breath:

• OpenAI GPT-4o: Pros are the insane speed and better vision capabilities for UI/UX stuff. Cons: sometimes it gets a bit lazy with code blocks compared to Sonnet.
• Google Gemini 1.5 Pro: Pros are that 2-million token context window which is basically a life-saver if you need to upload a whole repo at once. Cons: the reasoning feels slightly less "human" than what we're used to. Tbh, I just swap between them depending on the specific task. Waiting for a 2025 release date just feels like lost time when these are ready to go right now. Dont let your project stall while waiting for a version that might not even drop for another year!

Regarding what #11 said about "Been thinking about this since I saw the..." I'm gonna have to disagree a bit on the hybrid approach. Honestly, you might want to consider the long-term maintenance costs before you start mixing providers like that. It's basically a recipe for technical debt if you aren't careful. I've seen a lot of teams get burned by this lately. Here is why I'd be cautious:

• Model behavior changes over time so your "perfect" prompt today might break in six months
• Data privacy and security audits get way more complicated when your info is bouncing between different companies
• The complexity of managing multiple API keys and rate limits can kill your velocity Make sure to focus on building a robust abstraction layer that doesn't care which specific version you're using. Basically, dont wait for Claude 5 and dont rush into a messy multi-model setup either... just stick to a single stable environment until the landscape settles a bit. It's safer to have one system you understand perfectly than three you're constantly troubleshooting.

So, I actually went through a whole nightmare last year when I paused a dev project waiting for the 'next big thing' in LLMs, and tbh it was a total waste of time because the landscape shifted so fast that my original architecture became obsolete anyway. Respectfully, I'd consider another option besides just waiting for a specific release like Claude 5.

Hmm, I've had a different experience with these cycles—Anthropic is definitely unpredictable, and honestly, you might be waiting for a phantom date. Be careful about hitting the brakes now because:

- The jump from 3.5 to 4 or 5 usually involves massive architectural changes that might break your current implementation anyway
- You'll miss out on early-mover advantage while others are building on what's available today
- Rumors usually suggest we'll see more incremental 'Opus' or 'Sonnet' refreshes before a full version jump

I really think you should just build for the API you have now and focus on modularity so you can swap later. Waiting for a release that doesnt even have a leaked internal codename yet is risky. Just my two cents tho!

Like, I totally agree with the point about not letting ur budget get wrecked while waiting for the next big drop. If you look at what people are saying in the dev communities, the move is to stop worrying about specific dates and just optimize your spend right now. Honestly, instead of the standard sub, you should check out the Anthropic Claude API setup for your project. Tbh, it’s basically a game changer for keeping costs down if ur not hitting it 24/7. And you might want to look at Anthropic Claude 3.5 Haiku when you need to run bulk tasks because I think it’s meant to be way more cost-effective for iterations. Most reviews I’ve seen suggest that for long-term project setups, a pay-as-you-go approach saves a ton of money compared to a flat monthly fee, especially when the timeline for the next major version is so up in the air. It gives you the flexibility to pivot immediately the second the new tech lands without feeling like you overpaid for a billing cycle you didn't fully use