GPT-6.1 Sol is insane at browser tasks 🔥
> higher score than Astra
> 7.8x lower estimated cost
> highest score we’ve measured on Browser Use Bench 2.1
Why so cheap? 95% of its prompt tokens were cached and cached-token rate is 10x cheaper than Astra.
Opus 5.5 and Grok 4.7 also scored lower AND cost more.
11
1
125
10,606
@browser_use + @agentcardhq = magic
Browse with @browser_use, pay with @agentcardhq
We are happy to announce our new integration, now companies can navigate with Browser Use and pay with Agentcard.
2
3
24
4,395
Browser Use Ultrafast available 🔓
> 10x faster and cheaper
> compare flights for $0.004
> in <20s
Web agents are now fast, accurate, and cheap.
DM to skip the waitlist or comment why you want access 👀
Introducing: Browser Use Ultrafast, now on our cloud ⚡
10x faster and 10x cheaper.
> cheapest af
> extremely fast
> compare flights for $0.0041
Need more brain power? Also introducing BU Fast (smarter).
Compute is a bit limited, so we're letting people in batch by batch. DM me to jump the line 🔓
(1x speed btw)
16
13
167
23,601
Browser Use retweeted
Personal assistants cost $3k - $7k per user per year.
If you run headful computer for a whole year at $0.5/h (standard price), this is $3k per year or just raw VM compute...
From first principles you only need it:
> VM sandbox + bash
> control plane
> persistant filesystem
> headless browser
Browser Use VMs cost $175 PER YEAR ($0.02/h with full utilization).
That's 25x cheaper than raw headful computer use.
By first principles browser is the operating system for agents?
Hearing this week that personal assistants cost $3k-$7k per user per year, depending on scale effects / vertical integration / model selection. Implications/questions:
- Consumer subsidies will be a huge factor; people don’t care about privacy, they want utility and they want it for free.
- Model costs are deflating, but not fast enough for many startups. The dark horse is that Jev may rapidly deflate computer-use costs 100x.
- An interesting counter-position would be building an expensive and narrowly focused agent — perhaps combining proprietary supply (Carbone tables!) with a unique POV (“no one goes there anymore, go to The Grill instead”).
- Social is wide open but hinges on a key design challenge: how can agents increase social cohesion and not feel invasive in a group chat? A working social agent would rightly attract an avalanche of capital and be hard for incumbents to replicate.
- Finally, there are vertical focus areas like financial services where the LTVs outweigh the subsidy needed and are outside the near-term competitive focus of the incumbents. If it were me, I’d focus here — Credit Karma as an agent >> Credit Karma as an app.
Overall, it’s still very, very early, and we may very well be in the MySpace era of social or iPhone 2010. Hang onto your hats and keep building!!
33
8
190
25,514
Browser Use retweeted
New open-source MiMo V2.6 Flash nearly matches Grok for 1/57th the cost 🔥
Our browser benchmark:
> MiMo: 37.6
> Grok 4.7: 39.9
> Luna xhigh: 57.6
Still a long way to go. @XiaomiMiMo let's close that gap.
1
2
13
1,089
Browser Use Bench v2 Pareto frontier got completely redrawn today
> Claude Opus 5.5: 59.4
> GPT‑6 Sol medium: 66.9 (3.5x cheaper)
> GPT‑6 Luna xhigh: 57.6 (22x cheaper than Opus)
OpenAI is in its own league 🔥
All models available to try on our cloud.
[x-axis is log scale]
25
30
389
71,579
Open Source MiMo v2.6 models are the new Pareto frontier 🔥
> MiMo v2.6 Flash: 37.6
> Grok 4.7’s 39.9
> 57x lower recorded agent cost then Grok 4.7
> 27% cheaper then DeepSeek v4.1 Flash (low)
$0.10 vs $5.72 per task. 60 tasks each.
Open source models ❤️
7
4
69
8,239
Ultrafast reached 15k ⭐ in 4 days.
Breaking: Browser Use + Jev = Ultrafast ⚡
Findings flights took 7s and cost only $0.0039 🤯
> new action space every step
> DOM state space
> small LLM fallback to type
(this video is at 1x speed btw)
Built a tiny open source browser agent. try it below ↓
5
17
471
52,002
Grok 4.7 just dropped. Still chasing DeepSeek 👀
Long Horizon Browser Use Benchmark v2
> GPT-6 Astra: 80.6
> DeepSeek V4.1 Flash: 47.9
> Grok 4.7: 39.9
> Grok 4.6: 31.2
Better than 4.6. Still less than half Astra's score.
11
5
101
16,625
GLM 5.3 FlashX is getting really good, but clearly more expensive (also mogged by DeepSeek v4.1 Flash)
36
17
340
30,039
Browser Use retweeted
Very realistic overview. Hard to discern hype from reality about Jev so here are some bitter facts:
- Jev can’t write text: it’s not an autoregressive model, so forget about anything related to coding or writing
- Jev can’t see (yet): text only, althought there is nothing fundamentally preventing it from being multimodal. I’m sure they will add support for images?
- Jev can ONLY choose a set of predefined actions: the good thing about computer and browser use is that by very nature it’s a STATE -> ACTION model. The problem with long running state action models is that the reasoning and state understanding becomes extremely important.
- Jev is extremely cheap and fast
It opened up my eyes into what’s possible. We are often stuck in optimizing problems inside the box. This is one the real “think outside the box” solutions to problems we have been trying to solve.
I am extremely hyped about the future of computer and browser use. Latency matters, and people are clearly hyped about it.
We can surely combine some sort of global state understanding (LLM) with super fast actor model (System One). Obviously it’s possible - FSD and robotics companies have solved this already.
How hard can it be to apply the same thing to browser use?
18
11
100
16,683
ultra. fast. cloud.
ULTRAFAST is coming to Browser Use Cloud ⚡
> superhuman speed
> cheap
> undetectable
Tell us what you’d automate and get on the waitlist ↓
browser-use.com/ultrafast
7
5
139
17,209
Browser Use retweeted
ULTRAFAST is coming to Browser Use Cloud ⚡
> superhuman speed
> cheap
> undetectable
Tell us what you’d automate and get on the waitlist ↓
browser-use.com/ultrafast
Breaking: Browser Use + Jev = Ultrafast ⚡
Findings flights took 7s and cost only $0.0039 🤯
> new action space every step
> DOM state space
> small LLM fallback to type
(this video is at 1x speed btw)
Built a tiny open source browser agent. try it below ↓
18
17
352
61,981
Browser Use retweeted
114.8K ⭐ · Global Rank #100 · Browser Use by @browser_use
AI agents that can actually use the browser — navigate sites, click, type, and get things done. 🌐
Still adding ~1.1K stars this week, and now sitting right at GitHub’s Global Top 100. 🏅
star-history.com/browser-use…
2
6
728