We made Minimax M3.1 Free before and we are making it Unlimited now for our Pro and Ultra users.
Available now on your plans: perchai.app
Same as before, depending on how much support this post gets, we will turn it on for Starter too!
MiniMax M3.1 Flash is now available on Perch AI completely free of cost.
Our Free Lineup now includes:
1)MiniMax M3.1 Flash
2) GLM 5.3 Flash
3) DeepSeek V4.1 Flash
4) Beam 1.0
Try now: perchai.app
As promised! xitter.gq/PerchTerminal/status/2…
We made Minimax M3.1 Free before and we are making it Unlimited now for our Pro and Ultra users.
Available now on your plans: perchai.app
Same as before, depending on how much support this post gets, we will turn it on for Starter too!
2
51
The usage will not be metered against any of your limits, there's no 5hour limits or anything, just the 3 requests in flight to keep it fair to everyone.
For more queries: discord.gg/6ZFZQeHrw
3
88
Perch AI retweeted
We made Minimax M3.1 Free before and we are making it Unlimited now for our Pro and Ultra users.
Available now on your plans: perchai.app
Same as before, depending on how much support this post gets, we will turn it on for Starter too!
MiniMax M3.1 Flash is now available on Perch AI completely free of cost.
Our Free Lineup now includes:
1)MiniMax M3.1 Flash
2) GLM 5.3 Flash
3) DeepSeek V4.1 Flash
4) Beam 1.0
Try now: perchai.app
Should we? Let us know what model you would want on here or on our discord:
discord.gg/W7VQzybDak
We have extended the Unlimited DeepSeek Promo till the end of the weekend.
We have also doubled the limits for Free users.
Over the past 18 hours we have served Billions of DeepSeek tokens with speeds at 456 tok/s p90
Perchai.app
Replying to @Sammy026k
We do offer API per token rates to Enterprise users with committed spend for DS only. If you spend more than $1000/mo in tokens, happy to offer you an API.
29
Replying to @LOKPUMMXM @mdp_sec
That’s no longer the case, we have been self hosting for a bit now. The unlimited passes would be impossible on api lol.
Only time we use other providers now is for subs as a fallback less than 1% of our volume
5
38
Perch AI retweeted
Love to see us on the leaderboard! Good deals shouldn't have to come at the cost of weaker quality.
We aim to continue to provide the best service we can to y'all ♥️
15 deepseek hosts benched now
@PerchTerminal just joined at 5th after taking the challenge.
any other provider reckon their stack holds up?
dm access and ill run it
Love to see us on the leaderboard! Good deals shouldn't have to come at the cost of weaker quality.
We aim to continue to provide the best service we can to y'all ♥️
15 deepseek hosts benched now
@PerchTerminal just joined at 5th after taking the challenge.
any other provider reckon their stack holds up?
dm access and ill run it
Perch AI retweeted
The Perch CLI now has a full-screen view, in beta.
Your chats on the left, the conversation in the middle, agents, background tasks and changed files on the right. Click a tab or stay on the keyboard.
And /btw asks a quick question on the side while the work keeps going.
Type /tui to open it.
Replying to @1kartikkabadi1
Let us know how it goes! We intend to make this default once we can further smooth it out
3
38
Probably our biggest rewrite to date! Give us feedback so we can improve it further!
npm i -g perchai-cli, then /tui
Perchai.app/cli
The Perch CLI now has a full-screen view, in beta.
Your chats on the left, the conversation in the middle, agents, background tasks and changed files on the right. Click a tab or stay on the keyboard.
And /btw asks a quick question on the side while the work keeps going.
Type /tui to open it.
Replying to @Ausyeah23
Probably yes if all you want is cheap inference, providers with api might give you more value.
Our median speeds this week have been 250 tok/s+
1
71
Replying to @MiaAI_lab
This is not true, can personally assure you, there's no Google model coming this week or next week 🤣
1
3
166
Replying to @C4paGvahCEELqHc
It costs 29.99/week or 99.99/month! More details here perchai.app/docs/plans/unlim…
21
Replying to @Sammy026k
Tell that to our 100k+ users :) Ik you’re used to shitty harnesses where you need an API, we are not them.
If you want an api there’s as many shitty inference providers, we are not them either.
1
3
168
Replying to @mdp_sec @ClaudeDevs
Don’t see our name on their, benchmark it. We serve exclusively on AMD MI355x. We don’t mess around with quants. Data don’t lie.
xitter.gq/perchterminal/status/2…
As the promo comes to an end here's what we learned serving y'all Unlimited DeepSeek for 72 hours
We tested 5 ways to serve DeepSeek V4.1 Flash for agents.
The one that did the most total work felt the slowest: up to 80 s for a first token.
What we kept, where the time goes, and the network fix that halved slow first tokens 👇
2
3
158







