Corleth Labs

Independent and privacy respecting AI lab creating models for the AGI future.

[Corleth Labs @corlethlabs]

Our new video model, Corleth Motion, is now live in media studio!

Comment below for some free credits.

We made this video with the new media MCP.

[Corleth Labs @corlethlabs]

The dark horse prepares for war....

@TeamMcEntee

Created with Corleth

[Corleth Labs @corlethlabs]

We're putting aside some compute so we can get Qwen 4 support up ASAP once it drops.

[Uzi @uzairakrum]

Me showing my abliterated Qwen 4 27b to chatgpt and Claude users next week

[Corleth Labs @corlethlabs]

Corleth V2 is now out with a whole new interface and model catalog!

Comment below for some free credits.

https://t.co/mKmglNaZ5w

[Corleth Labs @corlethlabs]

🏳️‍🌈

[OrcaRouter 🐳 @OrcaRouter]

We’re taking uncensored model APIs off general public access.

We built them for research and security. Some users abused them for unlawful purposes. That crosses the line.

Going forward: KYC/KYB-verified companies and legitimate security research only.

Regular Qwen3.8 (model name: qwen/qwen3.8-27b) stays open: $0.33 input / $2.40 output per 1M tokens https://t.co/SD4GVlJxIR

Our mission remains the same: make AI infrastructure and security open, secure, and accessible. Security shouldn’t be a privilege reserved for the largest companies.

OrcaCyber launches next month. 🛡️
OrcaRouter users will be first.

Power to defenders. 🐳

[Corleth Labs @corlethlabs]

Dario is about to beat Elon on wealth once the IPO drops

[Corleth Labs @corlethlabs]

Unlimited Access to Qwen 3.5 9B should be a public funded resource. Take like 0.0000001% of the military budget.

[Corleth Labs @corlethlabs]

RIP @OpenAI

When this goes open source will be hosting this on Corleth.

[Kimi.ai @Kimi_Moonshot]

Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.

🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.

Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on https://t.co/85jA71X9gw, Kimi Work, Kimi Code, and the Kimi API.

Open weights by Nov 4, 2026.

🔗 API: https://t.co/XCrgjXAqMw
🔗 Tech blog: https://t.co/5iPP0MBej6

[Corleth Labs @corlethlabs]

What if we made Jev but it was slower and then it talked to you so you can understand it better. What if we also made it kind of bigger so it was smarter we could call it a large language model. Thoughts?

@typesafeai are you on this?

[Corleth Labs @corlethlabs]

The permanent non-AI underclass is already forming. You have maybe a few years to learn these tools and stay in that 0.2%.

Having frontier level AI models that are uncensored, privacy respecting, and market independent is an important step to giving the 0.2% access to 0.0006% flexibility.

[Corleth Labs @corlethlabs]

The full Orren 3 model set is now available to all paid users.

Along with this we have sent out a text and media reset.

If you make anything cool with these new models make sure to @ us!

[Corleth Labs @corlethlabs]

Orren 3 Spark is now live for all users. You can see it in your panel.

All stronger Orren 3 variants have been rolled out to a set of testing users. It will be available to all paid accounts on October 12th.

[Corleth Labs @corlethlabs]

Introducing Orren 3, Corleth's latest model series for coding and agentic work.

Orren 3 includes 4 variants at differing price, intelligence, and speed levels.

Orren 3 Spark - Nano model for everyday tasks.
Orren 3 Turbo - Frontier adjacent and cheap model.
Orren 3 - Our flagship model for tasks needing high intelligence.
Orren 3 Max - Orren 3 fine-tuned and running on our best hardware for the most demanding work.

Unlike Orren 2 all Orren 3 variants are 100% uncensored allowing you to work on your own terms.

Finally, Orren 3 supports High and Low reasoning levels, allowing you to change how deeply the model thinks depending on task difficulty.

You can read our full release blog post here: https://t.co/pdKI78uk0L

[Corleth Labs @corlethlabs]

Introducing Orren 3, Corleth's latest model series for coding and agentic work.

Orren 3 includes 4 variants at differing price, intelligence, and speed levels.

Orren 3 Spark - Nano model for everyday tasks.
Orren 3 Turbo - Frontier adjacent and cheap model.
Orren 3 - Our flagship model for tasks needing high intelligence.
Orren 3 Max - Orren 3 fine-tuned and running on our best hardware for the most demanding work.

Unlike Orren 2 all Orren 3 variants are 100% uncensored allowing you to work on your own terms.

Finally, Orren 3 supports High and Low reasoning levels, allowing you to change how deeply the model thinks depending on task difficulty.

You can read our full release blog post here: https://t.co/pdKI78uk0L

[Corleth Labs @corlethlabs]

At Corleth, there are zero restrictions on your AI chat use. Yell at it all you want!

[Andrew Curran @AndrewCurran_]

Effective November 12th, 2026, abusive behavior towards Claude will be a violation of Anthropic's Usage Policy.

[Qzoomer @Qzoomer1]

We have AI companies creating pure uncut Qino on a random Wednesday and you are nothing

[Corleth Labs @corlethlabs]

@Qzoomer1 It's not an image but yk wtv 🤣

[Corleth Labs @corlethlabs]

@Qzoomer1

[Corleth Labs @corlethlabs]

Orren 2 Prime Demo.

Prime gives you access to frontier level intelligence while respecting your privacy and autonomy.

Prime preview is available right now to $25 / month subscribers and will be officially released to all users next week.

[Corleth Labs @corlethlabs]

We're also giving everyone a usage reset for this update. You should see it reset within an hour.

[Corleth Labs @corlethlabs]

Our most intelligent model yet, Orren 2 Prime, is now available for preview in the CLI for $25 / month subscribers.

Orren 2 Prime is a 2.1T parameter MoE model that currently is benchmarking as high as Grok 4.7 and Qwen3.8 Max.

Share anything you're making with it and we'll repost you!

[Corleth Labs @corlethlabs]

Our most intelligent model yet, Orren 2 Prime, is now available for preview in the CLI for $25 / month subscribers.

Orren 2 Prime is a 2.1T parameter MoE model that currently is benchmarking as high as Grok 4.7 and Qwen3.8 Max.

Share anything you're making with it and we'll repost you!