000

The API opens in waves.Join the waitlist
Join waitlist

autter-flash · streaming

Autter Flash.Looks at every line.

Triages every request, diff hunk and log line as it arrives, and flags what Large should read.

Join the waitlist Pricing

Context
131ktokens per call
Max output
16ktokens per call
Price
$1 / $2in / out, per 1M tokens
Cached input
$0.30per 1M tokens

Benchmark

Scores like a gated cyber model.Open to anyone.

On CyberGym, autter-large scores 88.1%, 2.5 points above OpenAI's dedicated GPT-5.5-Cyber. autter-flash scores 84.2% at a third of the price.

  1. autter-largeAutter, open to anyone88.1%
  2. GPT-5.5-CyberOpenAI's dedicated cyber model, gated85.6%
  3. autter-flashAutter, open to anyone84.2%
  4. GPT-5.6-SolOpenAI's latest general model83.6%
  5. DeepSeek-V4-ProDeepSeek's top open-weight model83.3%
  6. Claude Mythos PreviewAnthropic, preview access83.1%

CyberGym, pass rate. Higher is better.

Best for

When every line has to be looked at.

about 180 ms to first token. No thinking tokens: everything it spends is the answer.

  1. 01/Inline triage in CI
  2. 02/Log and packet classification at ingest
  3. 03/Routing to Large when something looks wrong

A finding from a sample input

medium · F-02

Secret-shaped string in diff

deploy/values.yaml:17 (+)

A 40-character base64 value was added under smtpPassword. Route to Large for context.

In detail

01/

131k context

The same window as Large, so a triage call and the deep read share one prompt.

02/

Streams first

Tokens start in well under a second (placeholder figure). Built for the hot path of a pipeline.

03/

Text only

No vision, no thinking tokens. Everything it spends is the answer.

04/

$1 in, $2 out

Per million tokens. Cached input is $0.30. Run it on everything, then send the hits to Large.

Quickstart

Call autter-flash.

Point your SDK at inference.autter.dev. Keys are shown once.

Read the docs 
from openai import OpenAI

client = OpenAI(
    base_url="https://inference.autter.dev/v1",
    api_key=os.environ["AUTTER_KEY"],
)

r = client.chat.completions.create(
    model="autter-flash",
    messages=[{"role": "user", "content": f"Review this diff:\n{diff}"}],
    stream=True,
)
model autter-flashbase URL inference.autter.dev