Sébastien Conejo

CTO @ Manifest. Building the smart LLM Routing for OpenClaw agents. Berkeley SkyDeck Batch 21.

[Sébastien Conejo @SebConejo]

If you're paying $200/month for Claude Max 20x just to avoid the rate limit, there's a better way.

Plug @MiniMax_AI M3 behind your Claude Code through the Token Plan thanks to @Manifestforai. Same agent, 1.7B tokens/month, $20.

Full tutorial: https://t.co/4QjHOltAdV

[Sébastien Conejo @SebConejo]

Run Claude Code on Qwen models and stop hitting rate limits.

Or keep Claude models and let Qwen take over the second you hit your weekly cap. Same Claude Code, no more walls, top models for a fraction of the price.

I wWrote up the full setup here👇
https://t.co/NW8LC9QlVo

[Sébastien Conejo @SebConejo]

Yesterday @taranjeetio invited us to the @mem0ai offices to talk vision, strategy, and fundraising.

Mem0 builds the memory layer for AI agents, @Manifestforai build the routing layer.

Thank you Taranjeet for your insights and your time.

🦚https://t.co/Pm8n3zG6NN

[Sébastien Conejo @SebConejo]

@Bencera No I never signed up

[Sébastien Conejo @SebConejo]

Claude Code keeps stopping you mid-build? Here's the hack:
1. Go to https://t.co/28tIfkHydy
2. create a Claude Code agent
3. Ask your Claude code agent to update use Manifest as a provider.
4. route to any model: free, local, your subscription plans, custom models, @deepseek_ai, @ollama, etc...

Same agent loop, smaller bill.
https://t.co/STQQpti83x

[Sébastien Conejo @SebConejo]

@jahooma Cool stuff @jahooma, Bravo! I will play with it.

[Sébastien Conejo @SebConejo]

@Atenov_D Use Manifest, it's an open source router. You can route by complexity tiers, task-specific tiers and you can define custom routing tiers too -> https://t.co/cIJk4dfsh8

[Sébastien Conejo @SebConejo]

@sama I agree. cheaper would be nice though.

[Sébastien Conejo @SebConejo]

@artyomx @outsource_ If you want to scale, it will be the same. If you struggled with OpenClaw, yo uwill do with Hermes too. But it's still interesting to try it by yourself, see the differences, and make your own opinion

[Sébastien Conejo @SebConejo]

@huang_chao4969 @OpenRouter @xubinrencs Please create an SVG logo 😅. It could help with brand consistency.

[Sébastien Conejo @SebConejo]

@huang_chao4969 @OpenRouter @xubinrencs Love that. Well done guys 💪

[NVIDIA @nvidia]

2026: the year of explosive AI innovation.

@FireworksAI_HQ Co-founder & CEO Lin Qiao speaks on how the massive expansion of customized inference and models is creating speed of light production to scale timelines.

[Manifest @Manifestforai]

Manifest just broke into the top 100 apps on OpenRouter. #96 globally, 10.8B tokens routed in 30 days, 112 models in rotation. 💪
https://t.co/n4q4N5KQUW

[Manifest @Manifestforai]

10K+ pulls on Docker Hub 🎉
We pushed the Manifest image 2 weeks ago and it just crossed 10K downloads.
Thanks to everyone running it locally.

Today we support 16 cloud providers, including @ollama and @lmstudio for local setups.

What other local providers should we add?

[Sébastien Conejo @SebConejo]

Awesome Awesome Free LLM APIs 🤩
- 14 providers
- all with permanent free tiers (no trial credits)
- OpenAI SDK-compatible

━━━━━━━━━━━━
NATIVE PROVIDERS
━━━━━━━━━━━━
🇨🇦 Cohere
- Command A
- Command R+ / R / R7B
- Embed 4 (Text + Image)
+ 1 more: https://t.co/V1xFi1kI6v

🇺🇸 Google Gemini
- Gemini 2.5 Flash
- Gemini 2.5 Flash-Lite

🇫🇷 Mistral AI
- Mistral Small 4
- Mistral Medium 3
- Mistral Large 3
+ 3 more: https://t.co/YA9nMwmxAf

🇨🇳 Z AI
- GLM-4.7-Flash
- GLM-4.5-Flash
- GLM-4.6V-Flash

━━━━━━━━━━━━
INFERENCE PROVIDERS
━━━━━━━━━━━━

🇺🇸 Cerebras
- llama3.1-8b
- gpt-oss-120b
- qwen-3-235b
- zai-glm-4.7

🇺🇸 GitHub Models
- gpt-4.1
- gpt-4.1-mini
- gpt-4o
- o3-mini
+ 5 more: https://t.co/dfht7esl83

🇺🇸 Groq
- llama-3.3-70b-versatile
- llama-3.1-8b-instant
- llama-4-scout / maverick
- kimi-k2-instruct
+ 5 more

🇺🇸 Hugging Face
- Llama-3.1-8B
- Mistral-7B
- Mixtral-8x7B
+ 2 more: https://t.co/AfZwtTcaen

🇺🇸 Kilo Code
- grok-code-fast-1:free
- nemotron-3-super-120b:free
- trinity-large-thinking:free

🇬🇧 LLM7 io
- deepseek-r1-0528
- deepseek-v3-0324
- gemini-2.5-flash-lite
+ 3 more: https://t.co/4HgkWcEYqj

🇨🇳 ModelScope
- Qwen3.5-35B-A3B
- Qwen3.5-27B
- Qwen-Image

🇺🇸 NVIDIA NIM
- deepseek-r1
- nemotron-ultra-253b
- nemotron-3-super-120b
+ 5 more: https://t.co/UoDTnJ1HhK

🇺🇸 Ollama Cloud
- llama3.1:cloud
- deepseek-r1:cloud
- qwen2.5:cloud
+ 2 more: https://t.co/4UfehEyHVQ

🇺🇸 OpenRouter
- deepseek-r1-0528:free
- deepseek-chat-v3-0324:free
- qwen3.6-plus:free | 1M ctx
+ 7 more: https://t.co/DBTX3dro5E

🇨🇳 SiliconFlow
- Qwen3-8B
- DeepSeek-R1-Distill-Qwen-7B
- GLM-4-9B
+1 more: https://t.co/3NftJvRqET

If useful → RT 🙏
Spot a provider I missed? Drop it in the comments 👇

[Sébastien Conejo @SebConejo]

@TheGeorgePu This is a FAKE information

[Sébastien Conejo @SebConejo]

@TheGeorgePu I'm still viewing the check for pro plans on the pricing page.

[Sébastien Conejo @SebConejo]

@zeroclawlabs Beautiful work guys!

[Sébastien Conejo @SebConejo]

@AureaLibe It changes my schedule for the day every time. It’s really annoying