Bridgebench
@bridgebench
The official AI benchmark of the vibe coding movement @bridgemindai
11 Following 13.7K Followers
DeepSeek V4.1 Flash just dropped, and it's now the MOST EXPENSIVE small model on the BridgeBench lava lamp test.
DeepSeek V4.1 Flash: $0.04, 4m 3s
Claude Haiku 5.5: $0.03, 4m 34s
GPT 6 Luna: under $0.01, 42.8s
Luna built the best lamp of the three. Cheapest, fastest and best.
Also, DeepSeek's launch chart compares against Opus 5 and GPT 5.6 Sol. Two generations old.
GPT 6 Luna is still the best small model.
Show more
This week's NerfBench results are live. No nerfs detected.
Claude Opus 5.5: 105.8%
Claude Sonnet 5.5: 98.9%
GPT 6.1 Sol: 98.0%
GPT 6 Astra: 97.1%
The big story is Opus 5.5. It jumped 10 points in a week, from 95.7% to 105.8%.
Last week it was the only model below its launch score. This week it's the only one above it.
Opus 5.5 is getting stronger.
Show more
Gemini 4 Argon is now on BridgeBench.
Google put it up against GPT 6 Astra, Claude Fable 5.1, and Claude Opus 5.5 in their launch benchmarks.
On BridgeBench, it lost to all three. It lands at #8# with a 623.
It's behind Claude Sonnet 5.5, GPT 6 Sol, and even Claude Fable 5 from June.
Google's new flagship is a mid tier model.
Show more
Gemini 4 Argon is a solid improvement for Google.
But it comes nowhere near Claude Opus 5.5.
Disappointed that Google is still this far behind.
Gemini 4 Argon JUST DROPPED in the Antigravity CLI.
Is Google back???
Meta's new Muse Wonder is now on BridgeBench.
It barely beats Claude Haiku 5.5. Anthropic's smallest model.
Muse Wonder: 611
Claude Haiku 5.5: 601
It's only 5 points better than Meta's own Muse Spark 1.3, and 149 points behind Claude Opus 5.5.
Meta is still nowhere near the frontier.
Show more
Muse Wonder is benchmaxxed.
Meta says it beats the competition on the key benchmarks. On the BridgeBench lava lamp test it built a red cylinder.
No glass. No lava.
I've never seen a model this benchmaxxed in my life.
Show more
Latest NerfBench results are in. No nerfs detected.
Claude Opus 5.5: 95.7%
GPT 6 Astra: 102.0%
Claude Sonnet 5.5: 103.1%
GPT 6.1 Sol: 105.5%
All four are within normal variance.
But Opus 5.5 is the one to watch. It's the only model below its launch score.
We'll keep testing.
Show more
Claude Haiku 5.5 just dropped.
We put it head to head with GPT 6 Luna on the BridgeBench rocket launch test.
Claude Haiku 5.5: $0.05, 6m 58s
GPT 6 Luna: <$0.01, 1m 5s
Haiku built a clean scene, but Luna did it 6x faster for under a penny.
GPT 6 Luna is still the small model to beat.
Show more
Claude Haiku 5.5 just dropped and it completely fails the BridgeBench lava lamp test.
No glass. No lava. Just a yellow cylinder.
GPT 6 Luna built an actual lava lamp in 42.8 seconds for under a penny.
Haiku took 4m 34s and cost $0.03.
GPT 6 Luna is over 6x faster and 3x cheaper.
Show more
Fable 5.5 is coming.
Anthropic is insanely far ahead right now.
Fable 5.5 is dropping soon and it will be a massive leap.
Fable 5.5 is the model that slingshots them into their IPO.
I just cancelled 1 of my 4 Claude Max subscriptions.
Not because I don't like Opus 5.5. It's the best model I've ever used.
I cancelled because the usage is so good that I don't need 4 anymore.
OpenAI has to keep resetting limits to stop people from leaving. With Anthropic, great usage just comes with the subscription.
Show more
Mistral Large 4 just lost to DeepSeek V4.1 Flash.
Same sunset ocean prompt. Mistral rendered a blank screen. DeepSeek nailed it for 3 cents.
Claude Opus 5.5 and GPT 6 Astra built the best scenes of the four. Astra did it 3 minutes faster and for 30 cents less.
Europe's flagship model can't beat a Chinese flash model.
Show more
Mistral Large 4 just dropped and it completely fails the BridgeBench lava lamp test.
No lamp. No lava. Just a blown out light.
It is cheap though. $0.02 for the whole run.
GPT 6.1 Sol built a working lava lamp for $0.09.
Show more
OpenAI has to keep resetting usage and shipping improvements over the next 28 days for one reason.
To stop everyone from unsubscribing.
The models aren't good enough and the usage is horrible, so they're buying time with resets and patches.
Why not just focus on shipping a good model with good usage?
Show more
China has fallen behind in the AI race.
In July, Kimi K3 was supposed to be China's answer to American AI.
Today it couldn't even render a lava lamp.
Claude Sonnet 5.5 and GPT 6.1 Sol nailed it. Those are America's mid tier models.
Right now, China can't keep up with America's second best.
Show more
Anthropic gives their users way more compute than OpenAI.
GPT 6 Astra is practically unusable on a Codex subscription. I burn through my entire weekly limit on the $500 plan in a day.
GPT 6.1 Sol gives you a lot more usage. But Opus 5.5 is leagues ahead of it, and Anthropic's subscription still gives you way more.
Anthropic gives you a smarter model with more usage.
It is not even close right now.
Show more
Claude Opus 5.5 is miles ahead of every open source model.
Kimi K3 had everyone talking in July.
Where's China's answer to Opus 5.5?
Anthropic has done serious damage to OpenAI.
Opus 5.5 is insanely intelligent and affordable. GPT 6 Astra is less intelligent and costs 2.5x more.
And Fable 5.5 is dropping soon.
Even if GPT 6.1 Astra drops tomorrow, it's going to be hard to win me back.
A better model doesn't matter when your weekly limit runs out in a day.
Show more
