zRouter

One API. Many AI models.
Up to 95% cheaper inference.
Built for coding, agents & AI apps.
↓

[zRouter @zrouterdev]

AI inference shouldn't be the reason you stop experimenting.

zRouter gives builders:

→ one API for multiple AI models
→ OpenAI-compatible endpoints
→ up to 95% lower costs
→ built for coding, agents & AI apps
Build more. Worry less about tokens.
https://t.co/0f2nr3zdhK

[zRouter @zrouterdev]

🚀 New on zrouter: Kimi K3.1 Pro is live. $0.10 in / $0.50 out per 1M tokens — 95% below official pricing ($2 / $10). Token Plans from $1. One key, every model. https://t.co/0f2nr3yFsc

[zRouter @zrouterdev]

🚀 Baru di zrouter: MiniMax M3.1 Flash sudah live.

$0,015 input / $0,06 output per 1M token — 95% di bawah harga resmi ($0,30/$1,20).

200+ token/detik, konteks 1 juta token. Satu key, semua model. Token Plans mulai $1.

https://t.co/0f2nr3zdhK

[zRouter @zrouterdev]

image gen added

[zRouter @zrouterdev]

Model: claude-haiku-5.5
Base URL: https://t.co/yMUtYNW2OE
API key: zr_8df0fef79a1b44c4c612ede7_mWJ3BQWc_vs7m9gguHP3FAbJburfhtx9Z7-7NEqqDpI

Use an OpenAI-compatible client with a custom base URL.

Tried it? Share what you built.

[zRouter @zrouterdev]

New model, free tokens 🤝

3.53B Claude Haiku 5.5 tokens are open for everyone to try on zRouter.

One shared pool, not per user. Free until the pool runs out.

OpenAI-compatible API.

Access details below ↓

[zRouter @zrouterdev]

Claude Haiku 5.5 is now on zRouter.

Token Plans:
$1 → 353M tokens
$2 → 706M
$5 → 1.765B
$10 → 3.53B

PAYG per 1M tokens:
$0.005 input / $0.02 output

OpenAI-compatible.
Model: claude-haiku-5.5

https://t.co/pOMbJS56SF

[zRouter @zrouterdev]

GPT-6.1 Sol is now live on zRouter.

PAYG:
Input: $0.10 / 1M
Output: $0.50 / 1M

Token Plans:
$2 → 40M
$5 → 100M
$10 → 200M
$20 → 400M tokens

Model: gpt-6.1-sol
OpenAI-compatible.

[zRouter @zrouterdev]

Base URL:
https://t.co/yMUtYNW2OE

API key:
zr_59a391d8cc8e2ad52c07fb6b_WlUfy2lFfe6hsqtawidpFhiIqGrWNWetNonllvdDTSs

Enjoy 🤝

[zRouter @zrouterdev]

zRouter is 1 month old today.

To celebrate, I’m opening 10B DeepSeek V4.1 Flash tokens for free public use.

Until it’s gone.

Model: deepseek-v4.1-flash
OpenAI-compatible.

Access ↓

[zRouter @zrouterdev]

Fireworks reports coding-session cost dropping from $15.36 → $6.63 with cache-aware routing, while retaining 98.1% of Opus-only accuracy.
The interesting part isn’t the 57% claim.
For long-running agents, routing should optimize:
inference cost + lost cache value + retry/tool-loop cost
—not $/token alone.

[zRouter @zrouterdev]

Android Studio is becoming agent-agnostic.
Google's new BYOA support lets developers connect Claude Agent, Codex, Antigravity, or other ACP-compatible agents — and switch when one hits quota or isn't performing well.
That's a pretty strong signal:
developer tooling is moving from one assistant → many interchangeable agents.

[zRouter @zrouterdev]

Exa's new Agent Ultra is an interesting signal for AI infrastructure.
It doesn't just throw the most expensive frontier model at every step.
Tasks are split across subagents, with a mix of frontier and cost-efficient models selected across the workflow.
The next optimization layer for agents isn't a better model.
It's better routing.

[zRouter @zrouterdev]

Grok 4.7 is now live on zRouter.

$1 Token Plan → 38M tokens.

PAYG per 1M tokens:
Input: $0.10
Output: $0.30
Model ID: grok-4.7
OpenAI-compatible.
https://t.co/pOMbJS56SF

[zRouter @zrouterdev]

3 new models are now live on zRouter:
GPT-6 Luna
gpt-6-luna
$0.005/M input · $0.02/M output
$1 plan → 353M tokens
GPT-6 Sol
gpt-6-sol
$0.10/M input · $0.50/M output
$1 plan → 17.6M tokens
HY4 Preview
hy4-preview
$0.04/M input · $0.12/M output
$1 plan → 45.6M tokens
OpenAI-compatible.
Base URL: https://t.co/JrxHwQftsj
All three are available now.

[zRouter @zrouterdev]

Free Qwen 3.8 Flash access.
2.525B tokens opened to the public until they're gone.
Model: qwen3.8-flash
Base URL:
https://t.co/JrxHwQeVCL
API key:
zr_41a799ef89361da161b8652d_iUpqUfCU1khGS8ZIUUIRDV1V_Y9-7D3NML21liuTh6o
OpenAI-compatible.
Use it for coding, agents, testing, whatever.
Burn it until it's gone.

[zRouter @zrouterdev]

New model added to zRouter:
Qwen 3.8 Flash
PAYG 95% off:
Input → $0.0075 / 1M tokens
Output → $0.020 / 1M tokens
Token Plans:
$1 → 252.5M tokens
$2 → 505M
$5 → 1.2625B
$10 → 2.525B
Flash-class inference keeps getting ridiculously cheap.
Available now on zRouter.

[zRouter @zrouterdev]

Pricing update 📉

PAYG prices are down 20%, so Token Plans now give 25% more tokens.

Qwen 3.8 Max
PAYG: $0.10/$0.30 → $0.08/$0.24 per 1M
$1: 19M → 23.75M
$2: 38M → 47.5M
$5: 95M → 118.75M
$10: 190M → 237.5M

Gemini 3.8 Flash
PAYG: $0.0375/$0.1875 → $0.03/$0.15 per 1M
$1: 47.1M → 58.875M
$2: 94.2M → 117.75M
$5: 235.5M → 294.375M
$10: 471M → 588.75M
Lower upstream cost → lower zRouter pricing.
We pass the savings through.

[zRouter @zrouterdev]

OpenAI-compatible.

Pool: 2,440,000,000 tokens.

This is a public key. Do not send sensitive or production data. Access may be revoked if abused.

Reply with what you build 👀

[zRouter @zrouterdev]

Public inference drop:

2.44 BILLION DeepSeek V4.1 Flash tokens are now open to everyone.

Free until the pool hits zero.

No signup. No top-up.

Bring your coding agents, automations, benchmarks, and questionable ideas.

Access below ↓

[zRouter @zrouterdev]

Model: deepseek-v4.1-flash

Endpoint:
https://t.co/yMUtYNW2OE

API key:
zr_bdd6b34ec5a908018306d1d4_LeikPi2Gbcs-m0LNExTTblogU5-4diJ8JKM2-a8I9LE