Let your
agents run.

Qwen3.8-27B for the tools you already code with. Flat monthly pricing. No bill for every token.

Get your API key Free accounts: 0 requests/week.
A paid plan is required for API inference.

Ready for API inference? Paid plans from $19/month.

OMP, OpenCode, and your app connect through Yolo-Auto to Qwen3.8-27B for text, images, and tools.
Made for your stack
  • OMP
  • OpenClaw
  • Hermes
  • OpenCode
  • Cline
  • OpenAI SDK
  • 261.1Btokens served
  • 5.4Mrequests handled

Published benchmarks · Official Qwen model card

Qwen3.8-27B.
Coding benchmarks.

Compare Qwen3.8-27B with earlier Qwen models on three coding evaluations reported in the official model card. These are Qwen-reported results under the settings below, not independent evaluations or measurements of our API.

SWE-bench Pro

Agentic coding · Reported score

Qwen3.8-27B
61.7
Qwen3.6-27B
53.5
Qwen3.7-Plus
57.6

NL2Repo-Bench

Repo-level code generation · Reported score

Qwen3.8-27B
42.3
Qwen3.6-27B
36.2
Qwen3.7-Plus
41.1

DeepSWE 1.1

Agentic coding · Reported score

Qwen3.8-27B
42.2
Qwen3.6-27B
13.3
Qwen3.7-Plus
14.2

Charts by Yolo-Auto using the official Qwen3.8-27B model card, Text Performance table and evaluation notes, checked October 15, 2026. All three models use the Claude Code harness on these evaluations. SWE-bench Pro and DeepSWE 1.1 use temperature 1.0, top_p 0.95, and a 256K context window. For SWE-bench Pro, Qwen corrected problematic tasks and re-evaluated the baseline models on the refined benchmark.

For NL2Repo-Bench, Qwen disables Bash commands that attempt to access the specific repository, including pip download, pip install, and git clone. The cited table does not specify an evaluation date. These selected model results are not a benchmark of the Yolo-Auto endpoint or a claim of performance on every task.

Flat-rate.
By design.

Create an account, then choose a paid plan for API inference.

Paid subscriptions renew monthly. No token credits, no per-token overages. Cancel anytime.

Compare limits and estimated costs

Free

Create an account and choose a paid plan for API access.

$0
  • 0 requests/week
  • API requests require a paid plan
  • No card required to create an account
Create account

Builder

Best value

For coding agents and daily development.

$19/mo
  • No per-token billing
  • 256K context for repositories and long conversations
  • Approximately 200 million tokens a day*
  • Up to 5x concurrency
  • Cancel anytime
Choose Builder

Pro

For heavier interactive workloads that need more usage headroom.

$39/mo
  • Everything in Builder
  • 256K context for larger repositories
  • Approximately 750 million tokens a day*
  • Up to 10x concurrency
  • Cancel anytime
Choose Pro

Fleet

For sustained workloads that need three times Pro's usage allowance.

$99/mo
  • Everything in Pro
  • 3x Pro usage allowance*
  • 256K context for larger repositories
  • Up to 10x concurrency
  • Cancel anytime
Choose Fleet

*Usage limits assume an 87% cache hit rate.

Know where
your work goes.

Your work is not training data.

We process requests to generate responses, not to train models. Prompt and response bodies are not stored or retained.

We keep operational metadata such as model, token counts, status, and latency to run the service.

Read the privacy policy

Real infrastructure. Reachable people.

We operate a focused model-serving stack on dedicated capacity, with Cloudflare and other service providers supporting networking and operations.

Need help? Email hello@yolo-auto.com. You don’t need to join a Discord server to ask a question.

About Yolo-Auto

Go build.

Choose a paid plan for API inference.
Bring a task you actually want to finish.

Get your API key