Pareto Inference

GLM 5.3 Flash, DeepSeek V4.1 Flash and MiMo V2.6 Flash, served on our GPUs. Pay per token. Connect your coding agent, router, or app, or talk to us about white-glove, full-stack inference optimization.

Available on

Supported by

GLM 5.3 Flash

Uptime: 100.0%
TTFT: 0.87 s
TPS: 108.5
Cached input: 95.9%

DeepSeek V4.1 Flash

Uptime: 100.0%
TTFT: 0.79 s
TPS: 163.3
Cached input: 98.3%

MiMo V2.6 Flash

Uptime: 100.0%
TTFT: 1.28 s
TPS: 110.1
Cached input: 96.8%

Uptime history · all models12h

2026-10-28 10:42:51 UTC2026-10-28 22:42:51 UTC

Pricing

USD per 1M tokens · Limited-time prices
Model	Hours	Input	Output	Cached input
GLM 5.3 Flash60% under standard pricing		~~Standard price $0.15~~$0.06	~~Standard price $0.50~~$0.20	~~Standard price $0.03~~$0.012
DeepSeek V4.1 Flash30% under standard pricing	Off-peak	~~Standard price $0.15~~$0.105	~~Standard price $0.60~~$0.42	~~Standard price $0.003~~$0.0021
DeepSeek V4.1 Flash30% under standard pricing	Peak	~~Standard price $0.30~~$0.21	~~Standard price $1.20~~$0.84	~~Standard price $0.006~~$0.0042
MiMo V2.6 Flash10% under standard pricing		~~Standard price $0.14~~$0.126	~~Standard price $0.28~~$0.252	~~Standard price $0.0028~~$0.00252

Pricing details

Setup

Direct API

Coding agents

and more

Routers

and more

Chat apps

Install the OpenAI Python SDK and set PARETO_API_KEY in your environment.

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.paretoinference.com/v1",
    api_key=os.environ["PARETO_API_KEY"],
)

response = client.chat.completions.create(
    model="z-ai/glm-5.3-flash",
    messages=[{"role": "user", "content": "Hello, Pareto."}],
    stream=True,
)
for chunk in response:
    if chunk.choices and chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

Open setup guide

Install the OpenAI SDK and set PARETO_API_KEY in your server environment.

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.PARETO_API_KEY,
  baseURL: "https://api.paretoinference.com/v1",
});

const response = await client.chat.completions.create({
  model: "z-ai/glm-5.3-flash",
  messages: [{ role: "user", content: "Hello, Pareto." }],
  stream: true,
});

for await (const chunk of response) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}

Open setup guide

Set PARETO_API_KEY for OpenClaw and merge this into ~/.openclaw/openclaw.json. Start a new session.

{
  "agents": {
    "defaults": {
      "model": {
        "primary": "pareto/z-ai/glm-5.3-flash"
      }
    }
  },
  "models": {
    "mode": "merge",
    "providers": {
      "pareto": {
        "baseUrl": "https://api.paretoinference.com/v1",
        "apiKey": "${PARETO_API_KEY}",
        "api": "openai-completions",
        "models": [
          {
            "id": "z-ai/glm-5.3-flash",
            "name": "GLM 5.3 Flash (Pareto)"
          }
        ]
      }
    }
  }
}

Open setup guide

Set PARETO_API_KEY and merge this into ~/.config/opencode/opencode.json. Keep your existing providers and settings, then start OpenCode.

{
  "$schema": "https://opencode.ai/config.json",
  "model": "pareto/z-ai/glm-5.3-flash",
  "small_model": "pareto/z-ai/glm-5.3-flash",
  "provider": {
    "pareto": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Pareto Inference",
      "options": {
        "baseURL": "https://api.paretoinference.com/v1",
        "apiKey": "{env:PARETO_API_KEY}"
      },
      "models": {
        "z-ai/glm-5.3-flash": {
          "name": "GLM 5.3 Flash"
        }
      }
    }
  }
}

Open setup guide

Uses Claude Code Router. Back up ~/.claude/settings.json, save this as ~/.claude-code-router/config.json, and run ccr start. Set ANTHROPIC_MODEL and ANTHROPIC_DEFAULT_HAIKU_MODEL to pareto/z-ai/glm-5.3-flash.

{
  "HOST": "127.0.0.1",
  "PORT": 3456,
  "APIKEY": "your-local-ccr-key",
  "Providers": [
    {
      "name": "pareto",
      "api_base_url": "https://api.paretoinference.com/v1/chat/completions",
      "api_key": "$PARETO_API_KEY",
      "models": [
        "z-ai/glm-5.3-flash"
      ]
    }
  ]
}

Open setup guide

Uses LiteLLM. Run LiteLLM with this model, then add a Codex provider for http://127.0.0.1:4000/v1 with wire_api = "responses" and model pareto-flash.

model_list:
  - model_name: pareto-flash
    litellm_params:
      model: openai/z-ai/glm-5.3-flash
      api_base: https://api.paretoinference.com/v1
      api_key: os.environ/PARETO_API_KEY
      use_chat_completions_api: true
      additional_drop_params: ["web_search_options"]

Open setup guide

Set PARETO_API_KEY and merge this provider into ~/.pi/agent/models.json. Select it with /model.

{
  "providers": {
    "pareto": {
      "baseUrl": "https://api.paretoinference.com/v1",
      "api": "openai-completions",
      "apiKey": "$PARETO_API_KEY",
      "models": [
        {
          "id": "z-ai/glm-5.3-flash",
          "name": "GLM 5.3 Flash (Pareto)"
        }
      ]
    }
  }
}

Open setup guide

Set PARETO_API_KEY and save this as ~/.config/goose/custom_providers/pareto.json. Set GOOSE_PROVIDER=pareto and GOOSE_MODEL, then run Goose with the developer extension.

{
  "name": "pareto",
  "engine": "openai",
  "display_name": "Pareto Inference",
  "api_key_env": "PARETO_API_KEY",
  "base_url": "https://api.paretoinference.com/v1/chat/completions",
  "models": [
    {
      "name": "z-ai/glm-5.3-flash"
    }
  ],
  "requires_auth": true,
  "supports_streaming": true
}

Open setup guide

Set PARETO_API_KEY, then start Aider in your repository. Keep the openai/ prefix and ignore the model names Aider suggests.

export OPENAI_API_BASE=https://api.paretoinference.com/v1
export OPENAI_API_KEY="$PARETO_API_KEY"
aider --model openai/z-ai/glm-5.3-flash --no-show-model-warnings

Open setup guide

Put PARETO_API_KEY in ~/.continue/.env and merge this into ~/.continue/config.yaml. The editor extension does not read shell variables.

name: Pareto Inference
version: 1.0.0
schema: v1
models:
  - name: GLM 5.3 Flash (Pareto)
    provider: openai
    model: z-ai/glm-5.3-flash
    apiBase: https://api.paretoinference.com/v1
    apiKey: ${{ secrets.PARETO_API_KEY }}
    roles:
      - chat
      - edit
      - apply
    capabilities:
      - tool_use

Open setup guide

Set PARETO_API_KEY and merge this into ~/.omp/agent/models.yml. Start Oh My Pi with omp --model pareto/z-ai/glm-5.3-flash.

providers:
  pareto:
    baseUrl: https://api.paretoinference.com/v1
    api: openai-completions
    apiKey: PARETO_API_KEY
    authHeader: true
    models:
      - id: z-ai/glm-5.3-flash
        name: GLM 5.3 Flash (Pareto)

Open setup guide

Set PARETO_API_KEY and add this to ~/.config/crush/crushrc. Select pareto/z-ai/glm-5.3-flash in Crush. Use openai-compat, not openai.

provider add pareto --type openai-compat \
  --name "Pareto Inference" \
  --base-url "https://api.paretoinference.com/v1" \
  --api-key "$PARETO_API_KEY"

model add pareto/z-ai/glm-5.3-flash --name "GLM 5.3 Flash"

Open setup guide

Set PARETO_API_KEY, then start Qwen Code. Use --auth-type openai, not openai-responses.

export OPENAI_API_KEY="$PARETO_API_KEY"
export OPENAI_BASE_URL=https://api.paretoinference.com/v1
export OPENAI_MODEL=z-ai/glm-5.3-flash
qwen --auth-type openai

Open setup guide

Open Settings → Providers → Custom provider. Save Pareto and select the model. Ask Pareto for the current context and output limits; Kilo needs accurate limits for automatic context compaction.

Provider ID: pareto
Name: Pareto
API type: OpenAI Compatible
Base URL: https://api.paretoinference.com/v1
API key: Your Pareto API key
Model: z-ai/glm-5.3-flash

Open setup guide

Settings → Model settings → Add provider → Create custom provider. Use Chat completions, the Pareto base URL, and model z-ai/glm-5.3-flash. Clear Image, Video and PDF, and paste this into the model's Reasoning parameter mapping. The default mapping sends fields Pareto rejects.

{
  "reasoning_effort": reasoningLevel == "disabled" ? "none" : reasoningLevel == "enabled" ? "high" : reasoningLevel
}

Open setup guide

Add this model to your existing config. Keep your proxy access settings.

model_list:
  - model_name: pareto-flash
    litellm_params:
      model: openai/z-ai/glm-5.3-flash
      api_base: https://api.paretoinference.com/v1
      api_key: os.environ/PARETO_API_KEY

Open setup guide

Add this entry to config.yaml and restart CLIProxyAPI. The base URL keeps /v1. Apps use a key from api-keys, not your Pareto key.

openai-compatibility:
  - name: "pareto"
    base-url: "https://api.paretoinference.com/v1"
    api-key-entries:
      - api-key: "YOUR_PARETO_API_KEY"
    models:
      - name: "z-ai/glm-5.3-flash"

Open setup guide

Run the gateway from the Docker image; the npx build breaks streaming. Apps send your Pareto key and both x-portkey headers.

docker run -d -p 127.0.0.1:8787:8787 portkeyai/gateway:latest

curl http://localhost:8787/v1/chat/completions \
  -H "Authorization: Bearer $PARETO_API_KEY" \
  -H "Content-Type: application/json" \
  -H "x-portkey-provider: openai" \
  -H "x-portkey-custom-host: https://api.paretoinference.com/v1" \
  -d '{"model": "z-ai/glm-5.3-flash", "messages": [{"role": "user", "content": "Say hello."}], "stream": true}'

Open setup guide

Set PARETO_API_KEY and save this as config.json in the Bifrost app directory. The base URL has no /v1. Apps use the model pareto/z-ai/glm-5.3-flash.

{
  "providers": {
    "pareto": {
      "keys": [
        {
          "name": "pareto",
          "value": "env.PARETO_API_KEY",
          "models": [
            "*"
          ],
          "weight": 1
        }
      ],
      "network_config": {
        "base_url": "https://api.paretoinference.com",
        "default_request_timeout_in_seconds": 300
      },
      "custom_provider_config": {
        "base_provider_type": "openai",
        "allowed_requests": {
          "chat_completion": true,
          "chat_completion_stream": true,
          "list_models": true
        }
      }
    }
  }
}

Open setup guide

Run NextChat with your key on the server. BASE_URL has no /v1, and CUSTOM_MODELS adds the Pareto model.

docker run -d -p 3000:3000 \
  -e OPENAI_API_KEY="$PARETO_API_KEY" \
  -e BASE_URL="https://api.paretoinference.com" \
  -e CUSTOM_MODELS="-all,+z-ai/glm-5.3-flash@OpenAI" \
  -e DEFAULT_MODEL="z-ai/glm-5.3-flash@OpenAI" \
  yidadaa/chatgpt-next-web:v2.16.1

Open setup guide

Run AnythingLLM with the Generic OpenAI provider. Keep Max Tokens at 8192 or more; reasoning counts toward it.

docker run -d -p 3001:3001 --cap-add SYS_ADMIN \
  -v anythingllm_storage:/app/server/storage \
  -e STORAGE_DIR=/app/server/storage \
  -e LLM_PROVIDER=generic-openai \
  -e GENERIC_OPEN_AI_BASE_PATH=https://api.paretoinference.com/v1 \
  -e GENERIC_OPEN_AI_API_KEY="$PARETO_API_KEY" \
  -e GENERIC_OPEN_AI_MODEL_PREF=z-ai/glm-5.3-flash \
  -e GENERIC_OPEN_AI_MAX_TOKENS=8192 \
  mintplexlabs/anythingllm:1.16.2

Open setup guide

Add this endpoint to librechat.yaml, set PARETO_API_KEY in .env, and restart. Keep dropParams: Pareto rejects the user field.

endpoints:
  custom:
    - name: "Pareto"
      apiKey: ${PARETO_API_KEY}
      baseURL: "https://api.paretoinference.com/v1"
      models:
        default: ["z-ai/glm-5.3-flash"]
        fetch: true
      titleConvo: true
      titleModel: "current_model"
      dropParams: ["user"]

Open setup guide

Partners

Early access: New models live for partners before they open to everyone.
Invoicing and limits: Pay by invoice on a credit line, with limits set for your traffic.
Tuned to your use case: Shaped to your traffic. On coding workloads, about 95% of prompt tokens hit cache.
A direct line: A shared Slack channel and a dashboard for each model.

Frontier open-source model inference at up to a 60% discount

GLM 5.3 Flash

DeepSeek V4.1 Flash

MiMo V2.6 Flash