LiveKimi K3, GLM 5.2 & DeepSeek V4 — now live

Every open-source LLM.
One API.
Up to 70% cheaper.

Pixelit gives you one OpenAI-compatible API for 35+ open-weight models — DeepSeek, Llama, Qwen, Kimi, GLM & more — at 70% below official list price. No markup games, no vendor lock-in.

Get API Key Browse live pricing

Pay only for usageNo subscription requiredCancel anytime

35+
open models, one API key: 70%
below official list price: 100%
open-source weights: 1
API for every model

chat.py

1from openai import OpenAI
2
3# one key, every open-source model, 70% off
4client = OpenAI(
5    base_url="https://api.pixelit.pro/v1",
6    api_key="pxit-sk-7f3a...",   # from your dashboard
7)
8
9resp = client.chat.completions.create(
10    model="deepseek-v3.2",   # 70% below list price
11    messages=[{"role": "user", "content": "Hello, world"}],
12)
13print(resp.choices[0].message.content)

200 OK · 41ms · 1,204 tokens · $0.0004 — 70% cheaper than direct

Saved this month

$1,284.50

Smart routing

deepseek-v3.2 ✓ healthy

One key · Every open-weight lab · 35 models live

BBB

DDeepSeek

MLMeta Llama

AQAlibaba Qwen

MAMistral AI

MAMoonshot AI

ZAZhipu AI

MMiniMax

NNVIDIA

GGGoogle Gemma

AAArcee AI

GGroq

Live pricing

Open weights.
A fraction of the price.

List prices = official provider rates · you pay exactly 30%

ModelContextInput / 1MOutput / 1MYou save

Kimi K3

Input / 1M$3.00$0.90

Output / 1M$15.00$4.50

−70%

GLM 5.2

Input / 1M$0.76$0.23

Output / 1M$2.42$0.73

−70%

DeepSeek V4 Pro

Input / 1M$0.28$0.084

Output / 1M$0.42$0.13

−70%

DeepSeek V4 Flash

Input / 1M$0.090$0.027

Output / 1M$0.18$0.054

−70%

Kimi K2.7 Code

256K

Input / 1M$0.73$0.22

Output / 1M$3.50$1.05

−70%

MiniMax M3

Input / 1M$0.30$0.090

Output / 1M$1.20$0.36

−70%

Nemotron 3 Ultra

512K

Input / 1M$0.60$0.18

Output / 1M$3.60$1.08

−70%

Nemotron 3 Super

Input / 1M$0.085$0.025

Output / 1M$0.40$0.12

−70%

Mistral Medium 3.5

256K

Input / 1M$1.50$0.45

Output / 1M$7.50$2.25

−70%

Codestral

256K

Input / 1M$1.00$0.30

Output / 1M$3.00$0.90

−70%

gpt-oss-120b

131K

Input / 1M$0.037$0.011

Output / 1M$0.17$0.051

−70%

Trinity Large Thinking

256K

Input / 1M$0.22$0.066

Output / 1M$0.85$0.26

−70%

DeepSeek V3.2

DeepSeek

128K

Input / 1M$0.28$0.084

Output / 1M$0.42$0.13

−70%

DeepSeek R1

DeepSeek

128K

Input / 1M$0.28$0.084

Output / 1M$0.42$0.13

−70%

Llama 4 Maverick

Built for teams who ship
AI at scale

Everything you need to run production AI workloads — without the enterprise price tag or the infrastructure headaches.

One API, every open model

A single OpenAI-compatible endpoint for 35+ open-weight models. Swap DeepSeek for Llama or Qwen by changing one string — no new SDKs, no new keys.

Smart routing, honest errors

Requests route to the cheapest healthy deployment with automatic retries on transient blips. Real failures return clean, standard HTTP error codes — never silent degradation.

Up to 70% cheaper

We aggregate demand across thousands of teams and optimize infra, then pass the savings straight to you. Same models, same quality.

No rate-limit walls

Burst to thousands of RPM on day one. Our pooled capacity means you stop begging providers for quota increases.

Token-level analytics

See cost, latency and usage per model, per key, per feature. Export everything. Know exactly where every dollar goes.

Enterprise-grade trust

SOC 2 Type II, zero data retention mode, SSO/SAML, and region pinning. Your prompts are never used for training.

Developer experience

Migrate in 30 seconds.
Change one line.

Pixelit speaks the OpenAI API fluently. Point your existing code at our endpoint and instantly unlock every open-weight model — at 70% off list price.

Drop-in replacement — keep your existing OpenAI SDK code
Switch models by changing one string, not your whole stack
Automatic retries, clean standard errors, load balancing built in

Read the full API docs

main.py

1from openai import OpenAI
2
3client = OpenAI(
4    base_url="https://api.pixelit.pro/v1",
5    api_key=os.environ["PIXELIT_API_KEY"],
6)
7
8resp = client.chat.completions.create(
9    model="deepseek-v3.2",
10    messages=[{"role": "user", "content": "Hello!"}],
11)

OpenAI SDK compatible · streaming, tools & vision supported

Coding agents

Works inside your coding agent

Claude Code, Codex, Cursor, opencode, pi, Aider — if your tool speaks OpenAI or Anthropic, it speaks Pixelit. Cheap open models inside your editor in under a minute.

Claude Code

2 env vars and you're in

export ANTHROPIC_BASE_URL=https://api.pixelit.pro
export ANTHROPIC_AUTH_TOKEN=pxit-sk-...

Codex CLI

~/.codex/config.toml

[model_providers.pixelit]
base_url = "https://api.pixelit.pro/v1"

Cursor

Settings → Models

Override OpenAI Base URL:
https://api.pixelit.pro/v1

opencode

opencode.json provider

"pixelit": {
  "npm": "@ai-sdk/openai-compatible"

pi

one tiny extension

pi.registerProvider("pixelit", {
  baseUrl: "https://api.pixelit.pro/v1",

Aider

env vars, any model

export OPENAI_API_BASE=https://api.pixelit.pro/v1
aider --model kimi-k3

Full setup guides for every agent

Pricing

Pay for tokens, not seats

Choose pay-as-you-go, four-day Kimi/GLM access, or the complete 33-model subscription.

Pay As You Go

$0no subscription

Top up credits and use any of the 33 models at token-based rates.

No subscription required
Credit packs from $10 — never expire
All 33 models available
Pay only for tokens used
Smart routing and clean errors

Get API key

4-Day Unlimited

$40/ 4 days

Unlimited access to kimi-k3 and glm-5.2; renews every four days.

Renews automatically every 4 days
Unlimited Kimi K3 and GLM 5.2 usage
Priority routing queue
120 RPM burst rate limit
Priority email support

Choose plan

Full Access 4K

$50/ month

4,000 requests every day across all 33 models.

4,000 requests every day
All 33 models included
Priority routing queue
120 RPM burst rate limit
Priority email support

Coming soon

Pay as you go covers all 33 models · The 4-day plan is limited to Kimi K3 and GLM 5.2

Test checkout will be enabled after the private Dodo credentials are installed.

Loved by builders

Teams are saving big with Pixelit

From indie hackers to production teams — builders route their open-model traffic through Pixelit.

“We cut our inference bill from $14k to $3.8k a month by moving to Pixelit. The migration literally took one afternoon.”

Sarah Chen

CTO, Looma

“Their honest error model saved our launch — clean 429s when we hit limits, instant retries on blips, zero silent failures to debug.”

Marcus Webb

Founder, Draftly

“One key for every open model is a superpower. We A/B test DeepSeek vs Llama vs Qwen in production by changing a query param.”

Priya Nair

AI Lead, Hexlab

No subscription required · Top up when ready

Start building with every open model today

Grab your API key and make your first discounted call in under two minutes. Your wallet will thank you.

Get your API key See pricing

pxit-sk-… · works with OpenAI SDK, LangChain, Vercel AI SDK, LlamaIndex

Every open-source LLM.One API.Up to 70% cheaper.

Open weights.A fraction of the price.

Built for teams who shipAI at scale

One API, every open model

Smart routing, honest errors

Up to 70% cheaper

No rate-limit walls

Token-level analytics

Enterprise-grade trust

Migrate in 30 seconds.Change one line.

Works inside your coding agent

Claude Code

Codex CLI

Cursor

opencode

pi

Aider

Pay for tokens, not seats

Pay As You Go

4-Day Unlimited

Full Access 4K

Teams are saving big with Pixelit

Start building with every open model today

Every open-source LLM.
One API.
Up to 70% cheaper.

Open weights.
A fraction of the price.

Built for teams who ship
AI at scale

Migrate in 30 seconds.
Change one line.