Stop picking modelsGo modeless

A new model takes the lead every few weeks, and keeping up is nobody's job. Point the agent you already use at Modeless and it makes that choice for you while you work.

Get started Set up your agent

No credit card · OpenAI-compatible · Live in 60 seconds

30^%

Auto lower cost per task than Claude Opus 5.5

SWE Pro, DeepSWE v1.1, Terminal-Bench 4.0

10^%

Fusion higher score than Claude Opus 5.5

DRACO

50^%

Fusion lower cost per task than Claude Opus 5.5

DRACO

01Set up

Works in the agent you already use

Pick your tool and copy two or three settings. You keep the editor and the workflow you already have.

All setup guides and API basics

Anthropic Messagesrecommended: auto

Create an API key
Sign up at modeless.ai and create a key in the console.

Point Claude Code at Modeless

Add these to your shell profile, then open a new terminal.

bash

export ANTHROPIC_BASE_URL="https://api.modeless.ai"
export ANTHROPIC_AUTH_TOKEN="<your-modeless-key>"
export ANTHROPIC_MODEL="auto"

✓
Check it works
Run claude and send one message. /status shows the Modeless base URL.
- · Routing survives context compaction: the session keeps its model.
- · Sub-agents get their own routing state, separate from the parent.

OpenAI Responsesrecommended: auto

Create an API key
Sign up at modeless.ai and create a key in the console.

Add a provider to ~/.codex/config.toml

toml

model = "auto"
model_provider = "modeless"

[model_providers.modeless]
name = "Modeless"
base_url = "https://api.modeless.ai/v1"
env_key = "MODELESS_API_KEY"
wire_api = "responses"

Export your key

bash

export MODELESS_API_KEY="<your-modeless-key>"

Check it works
Run codex and send one message. The header shows model auto.

OpenAI Chat Completionsrecommended: auto

Create an API key
Sign up at modeless.ai and create a key in the console.

Add Modeless to opencode.json

json

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "modeless": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Modeless",
      "options": {
        "baseURL": "https://api.modeless.ai/v1",
        "apiKey": "{env:MODELESS_API_KEY}"
      },
      "models": {
        "auto": { "name": "Auto" },
        "fusion": { "name": "Fusion" }
      }
    }
  },
  "model": "modeless/auto"
}

Check it works
Run opencode, pick modeless/auto with /models, and send one message.

OpenAI Chat Completionsrecommended: auto

Create an API key
Sign up at modeless.ai and create a key in the console.

Open Cline settings

Set API Provider to OpenAI Compatible, then fill in:

text

Base URL   https://api.modeless.ai/v1
API Key    <your-modeless-key>
Model ID   auto

Check it works
Start a new task. The task header shows model auto.

OpenAI Chat Completionsrecommended: fusion

Create an API key
Sign up at modeless.ai and create a key in the console.

Settings → Models

Enable Override OpenAI Base URL, paste your key, and add custom models:

text

OpenAI Base URL   https://api.modeless.ai/v1
OpenAI API Key    <your-modeless-key>
Custom models     auto, fusion

Check it works
Select auto in the chat model picker and send one message.

02The models

Two ways to use many models

Spend less

Auto`model: "auto"`

One model string picks the right model for every request, so you stop paying frontier prices for easy work.

Best for: Long coding-agent sessions where most turns are routine.

How Auto works

Answer better

Fusion`model: "fusion"`

Several models answer in parallel. One model reads them all and writes the reply.

Best for: Design reviews, hard questions and plans, where the answer has to be right.

How Fusion works

03Benchmarks

Measured on cost and score together

Each chart plots cost per task against score, with single models run on the same harness for comparison. A lower price only counts if the score holds up.

Auto`model: "auto"`

Agent: TBD · Tasks: TBD · Runs per configuration: TBD · Cost: measured tokens × public USD prices

Auto results and method

Fusion`model: "fusion"`

Harness: TBD · Questions: TBD · Runs per configuration: TBD · Cost: every panel and aggregator call, measured tokens × public USD prices

Fusion results and method

04Why this exists

Nobody has time to keep
picking the right model

A new model takes the lead every few weeks, on one benchmark, for one kind of task. Checking which one to use is a job nobody was hired to do.

So most people pick the strongest model and pay for it on every turn, including the turns where it reads a file or lists a directory. Modeless makes that choice for you while the work goes on.

05Pricing

Only model cost

+5%

Model cost: $1.00
Modeless fee (5%): $0.05
You pay: $1.05

on top of what the models cost. Pay as you go. Your console shows every request and the model that served it.

How billing works

06Questions

Before you switch

All questions

Try it on a real workload

Change two settings in the agent you already use and see what it costs on your own work.

Get started Set up your agent

Stop picking modelsGo modeless

Results

Works in the agent you already use

01Claude CodeAnthropic Messages

Create an API key

Point Claude Code at Modeless

Check it works

02Codex CLIOpenAI Responses

Create an API key

Add a provider to ~/.codex/config.toml

Export your key

Check it works

03OpenCodeOpenAI Chat Completions

Create an API key

Add Modeless to opencode.json

Check it works

04ClineOpenAI Chat Completions

Create an API key

Open Cline settings

Check it works

05CursorOpenAI Chat Completions

Create an API key

Settings → Models

Check it works

Two ways to use many models

Automodel: "auto"

Fusionmodel: "fusion"

Measured on cost and score together

Automodel: "auto"

Fusionmodel: "fusion"

Nobody has time to keep picking the right model

Only model cost

Before you switch

01Is Modeless right for every task?

02How should I evaluate it?

03Will Auto switch models mid-conversation? Will my cache break?

04What happens if a model fails after streaming starts?

Try it on a real workload

Auto`model: "auto"`

Fusion`model: "fusion"`

Auto`model: "auto"`

Fusion`model: "fusion"`

Nobody has time to keep
picking the right model