Pinned Tweet
Claude Code is being carelessly wasteful with your tokens.
By our benchmarking, it uses ~2x more tokens than needed. Half your limit, gone by default.
Brad Eckert retweeted
OPUS 5 first results:
WOZCODE vs Claude Code — fixing bugs in an Express + TypeScript API (6 real prompts):
💰 $13 vs $21 (41% cheaper)
🔁 96 vs 227 turns (58% fewer)
⏱️ 31m 11s vs 36m 44s (15% less time)
⭐ quality 8.1 vs 8.1 (of 10)
Full numbers: wozcode.com/benchmark-histor…
Kimi K3 + WOZCODE is higher quality and 72% cheaper than Fable 5 without WOZCODE
Kimi K3 BEATS Claude Code on our benchmarks.
6 real bug-fix & feature tasks on an Express + TypeScript API, blind-judged:
🏆 Quality: 8.1 vs 7.8 — the open-weights model outscores stock Claude Code running Claude Fable 5
💰 Cost: $8.69 vs $31.19 — 72% cheaper
🔁 Turns: 101 vs 174
⏱️ The tradeoff: ~2× slower (89m vs 46m)
Full numbers: wozcode.com/benchmark-histor…
Brad Eckert retweeted
GPT5.6 Sol
WOZCODE (in Codex) vs Codex — fixing bugs in an Express + TypeScript API (6 real prompts):
💰 $4.74 vs $8.50 (44% cheaper)
🔁 18 vs 23 turns (22% fewer)
⏱️ 6m 58s vs 14m 50s (53% less time)
⭐ quality 7.8 vs 7.5 (of 10)
Full numbers: wozcode.com/benchmark-histor…
Brad Eckert retweeted
FABLE 5:
WOZCODE vs Claude Code — screenshot-driven art portfolio builds (10 real prompts):
💰 $13 vs $31 (57% cheaper)
🔁 42 vs 202 turns (79% fewer)
⏱️ 31m 7s vs 45m 14s (31% less time)
⭐ quality 8.1 vs 7.3 (of 10)
Full numbers: wozcode.com/benchmark-histor…
Fable 5 update --> WOZCODE saves you even more with better models
FABLE 5:
WOZCODE vs Claude Code — fixing bugs in an Express + TypeScript API (6 real prompts):
💰 $15 vs $33 (56% cheaper)
🔁 51 vs 240 turns (79% fewer)
⏱️ 27m 59s vs 44m 24s (37% less time)
⭐ quality 8.1 vs 7.7 (of 10)
Full numbers: wozcode.com/benchmark-histor…
Brad Eckert retweeted
Went off the coast of California with a fish killing robot from @ShinkeiSystems yesterday.
The robot kills the fish using the Japanese Ikejime technique where the brain is spiked and the gills are slit to bleed the fish. It all sounds a bit rough, but it's this versus the fish suffocating on deck for hours and being stressed.
The result of the Ikejime process is much higher grade fish. Also we saw whales.
Brad Eckert retweeted
Can't make the case for WOZCODE any better than this.
Companies need to be able to control costs while also leveraging frontier models
Our Anthropic bill is about to jump from $400K → $1.4M/yr.
Not because usage exploded, but because we're about to cross 150 seats.
Past 150 seats you're forced into Enterprise tier. Seats stop including any usage, every token bills at standard API rates. At our current run rate that's 3.5x overnight.
Unfiltered thoughts on AI spend:
1. We should spend tokens to grow as aggressively as possible. But most people (me included) aren't conscious of what they're spending.
2. Visibility comes first. People see their personal number and they're shocked. I accidentally spent $4,000 in 3 days in Claude Code.
3. For engineering the spend is clearly worth it. Pay for the best model, it saves more than it costs.
4. For a lot of other roles it's questionable. Apps nobody uses, skills someone already built. No ROI.
5. Spend limits are coming. We already require approval for more tokens on our support team.
The era of token-maxxing is coming to an end.
Brad Eckert retweeted
Claude Code is being carelessly wasteful with your tokens.
By our benchmarking, it uses ~2x more tokens than needed. Half your limit, gone by default.
Brad Eckert retweeted
WOZCODE reads files differently.
Instead of dumping entire file contents into context, AST truncation replaces function bodies with stubs. Types, exports, and signatures stay intact.
Same useful information. 40 to 60% fewer tokens.
And because every token gets re-ingested on every subsequent call, a leaner read compounds across the entire session.
Built in. No configuration.
AI can waste all the water in the world as long as it keeps making dumb pictures of my dog to send my fiancée every time she asks me to do something around the house
Stop tokennmaxxing
The entire vibe coding wave got one thing backwards
Everyone optimized for speed
Prompt in, app out, as fast as possible
But the bottleneck isn't just speed
It's context
Give coding agents better context and everything improves.
Lower cost
Faster execution
Better output
@WithWoz1 is built for that shift.
A Claude Code plugin that makes coding agents faster, cheaper, and higher performing
Up to 55% cheaper
Up to 40% faster
Higher benchmark scores than Claude Code alone
And the timing matters:
The age of tokenmaxxing is fading.
The age of more measurable ROI is emerging.
🎙️ Ben Collins (@BenjaminT007), CEO & Co-Founder, Woz (@WithWoz1) on @fondocom @thestartpod
Huge thanks to @EragonAI for making this episode possible!
Brad Eckert retweeted
The entire vibe coding wave got one thing backwards
Everyone optimized for speed
Prompt in, app out, as fast as possible
But the bottleneck isn't just speed
It's context
Give coding agents better context and everything improves.
Lower cost
Faster execution
Better output
@WithWoz1 is built for that shift.
A Claude Code plugin that makes coding agents faster, cheaper, and higher performing
Up to 55% cheaper
Up to 40% faster
Higher benchmark scores than Claude Code alone
And the timing matters:
The age of tokenmaxxing is fading.
The age of more measurable ROI is emerging.
🎙️ Ben Collins (@BenjaminT007), CEO & Co-Founder, Woz (@WithWoz1) on @fondocom @thestartpod
Huge thanks to @EragonAI for making this episode possible!
Brad Eckert retweeted
Claude Code makes 12 API calls for a single refactor task.
While WOZCODE makes just 2.
Choose wisely
Want to understand why WOZCODE is faster than vanilla Claude Code?
It comes down to how many API calls your agent actually makes.
Most Claude Code agents treat every action as a separate roundtrip. Find files. Filter them. Read them. Edit each one. Verify each edit. A single refactor task can hit 12 calls before you know it.
WOZCODE has two tools that change this completely.
Smart Search combines file discovery, content filtering, and reading into one call. Instead of running Glob, then Grep, then Read on each file separately, it finds your files and returns their contents in a single response.
Batch Edit stacks all your file changes into one call. Instead of editing files one at a time, you pass an edits array and every change lands in a single roundtrip.
The same task that takes 12 calls in vanilla Claude Code takes 2 in WOZCODE.
That difference compounds across every session. Fewer roundtrips means smaller context on every subsequent call. Lower token usage. Lower cost. Faster execution.
Both tools are built into every WOZCODE session. No configuration. No setup. They just work.
That is how 10x fewer roundtrips happens in practice. Not from a rewrite. Not from a new model. Just from batching what was already being done one call at a time.
Brad Eckert retweeted
Want to understand why WOZCODE is faster than vanilla Claude Code?
It comes down to how many API calls your agent actually makes.
Most Claude Code agents treat every action as a separate roundtrip. Find files. Filter them. Read them. Edit each one. Verify each edit. A single refactor task can hit 12 calls before you know it.
WOZCODE has two tools that change this completely.
Smart Search combines file discovery, content filtering, and reading into one call. Instead of running Glob, then Grep, then Read on each file separately, it finds your files and returns their contents in a single response.
Batch Edit stacks all your file changes into one call. Instead of editing files one at a time, you pass an edits array and every change lands in a single roundtrip.
The same task that takes 12 calls in vanilla Claude Code takes 2 in WOZCODE.
That difference compounds across every session. Fewer roundtrips means smaller context on every subsequent call. Lower token usage. Lower cost. Faster execution.
Both tools are built into every WOZCODE session. No configuration. No setup. They just work.
That is how 10x fewer roundtrips happens in practice. Not from a rewrite. Not from a new model. Just from batching what was already being done one call at a time.
Brad Eckert retweeted
if you really want to show Claude Code what's up you'd support WOZCODE from @WithWoz1 in Codex. Codex+ WOZCODE dominates
Brad Eckert retweeted
Did you know WOZCODE tracks every dollar it saves you?
Run /woz-savings and get the full breakdown:
📊 Session Report
→ Calls made vs. vanilla Claude Code
→ Tokens saved this session
→ Exact dollars back in your pocket
📈 Lifetime Report
→ Total calls batched across every session
→ Hours of roundtrip time eliminated
→ Cumulative $ saved since day one
All pulled from your real local usage not estimates. The numbers in the video? A live savings report straight from an actual session.
How the math works:
1. WOZCODE logs every batched call
2. Compares to what vanilla Claude Code would have used
3. Tokens × rate = your real savings
Don't take our word for it, prove it to yourself.
See how much WOZCODE can save you
Want to see how much WozCode can save on your own repo?
You don’t need a demo project.
You can run a benchmark directly on an existing codebase and compare the difference between Claude Code and WozCode.
The flow is simple:
Choose a repo you already work with
Run the same development task
Compare Claude Code vs WozCode side by side
See the cost difference and savings clearly
WozCode shows how much was spent, how many turns it took, and how much could be saved when the same work is executed more efficiently.
This makes it easier for teams to understand the real cost of their AI coding workflows.
Not in theory.
Not from a generic benchmark.
But from their own repo, their own tasks, and their own usage patterns.
If your team is using Claude Code regularly, benchmarking your repo is the fastest way to see where tokens are going and how much WozCode can help reduce the cost.
Run a benchmark.
Compare the difference.
See the savings.
Brad Eckert retweeted
i was in sf last week and finally visited the @WithWoz1 office and met the team @BenjaminT007, @ChefEckert irl. it was really fun being in the same room with everyone after working remotely. the team is super experienced, sharp, and moves fast. one night we were at the hotel talking about a feature late at night, and the next day we had already started building it.
i’ve also been getting more hands on with wozcode. it’s fun, challenging, and lowkey addictive.
we crashed @AnthropicAI code with claude event with WOZCODE signs.
i visited replit hq and went to @vibecodeapp_ hq met @anshnanda finally, loved the space and the vibe there. we went for a walk and talked about where things are headed.
the energy in sf is something else. i was at @UseCorgi late at night building with people without even checking the time, and it just felt right.
every time i visit, it makes me wanna move there a little more. maybe soon…
Brad Eckert retweeted
WOZCODE is live on Product Hunt 🚀
Cut your Claude cost by up to 50%
Check it out:
producthunt.com/products/woz…
