basetenlabs/Hack-the-North-2026

basetenlabs/Hack-the-North-2026

Stars: 0

Forks: 0

Open issues: 0

Created: 2026-08-18T01:50:30Z

Pushed: 2026-09-09T15:40:15Z

Default branch: main

Fork: no

Archived: no

README:

Hack the North 2026 x Baseten

Hey hackers!

We're excited to meet everyone at Hack the North this weekend. This guide covers the prizes, credits, hosted models, coding tools, and H100 training access available through Baseten during the event.

What is Baseten?

Baseten is an inference-focused AI infrastructure platform used by AI companies including Descript, Bland, Clay, Writer, and OpenEvidence.

For this hackathon, you can use Baseten in three ways:

• Call hosted open-source models such as DeepSeek, Kimi, GLM, Inkling, GPT-OSS, and Nemotron

through a single OpenAI-compatible API.

• Use H100 GPU workstations to train or fine-tune your own models.
• Route Claude Code, Codex CLI, or Pi through Baseten-hosted models with Baseten Switch.

You do not need to manage inference servers or GPU infrastructure. Start with the hosted Model APIs for the quickest path to a demo, use Baseten Switch to power your coding agent, or visit the Baseten booth if your project needs training compute.

Prizes

Grand Prize

• A free trip to San Francisco to meet the Baseten team
• Final-round interviews with Baseten
• $200 in Baseten credits
• A signed hardcover copy of *Inference Engineering*

Runner-Up

• $100 in Baseten credits
• AirPods 4
• A signed hardcover copy of *Inference Engineering*

Credits

Every hacker can claim free Baseten credits with promo code shared in the event Slack! Check #spons-baseten-2026 for more details!

To redeem the code:

1. Create a Baseten account. 2. In the Baseten dashboard, select Billing and usage in the left sidebar. 3. Scroll to Redeem promo credits. 4. Enter the promo code and redeem it.

Credits apply to the workspace, not each individual member. If your team shares one Baseten workspace, the person who created that workspace must redeem the code. You only need to redeem it once for the whole team.

Inference

Model APIs let you call hosted LLMs with the OpenAI Chat Completions API or the Anthropic Messages API. Baseten manages the model deployment and serving infrastructure.

Get started

1. Create an API key. 2. Save it in your shell. Do not commit it to Git. 3. Install the OpenAI Python SDK. 4. Point the SDK at Baseten and choose a model from the live catalog.

export BASETEN_API_KEY="your-api-key"
pip install openai

import os

from openai import OpenAI

client = OpenAI(
api_key=os.environ["BASETEN_API_KEY"],
base_url="https://inference.baseten.co/v1",
)

response = client.chat.completions.create(
model="zai-org/GLM-5.3",
messages=[
{"role": "user", "content": "Pitch me a hackathon project."},
],
)

print(response.choices[0].message.content)

The API also supports streaming, tool calling, structured outputs, JSON mode, reasoning, vision, and audio. Support varies by model, so check the live documentation before choosing one.

Useful links:

• Model APIs overview and quickstart
• Current models, features, context windows, and model slugs
• Pricing, rate limits, and prompt caching
• Tool calling
• Structured outputs
• Reasoning
• Vision
• Audio

You can also query the live model catalog from your terminal:

curl https://inference.baseten.co/v1/models \
-H "Authorization: Bearer $BASETEN_API_KEY"

Rate Limits

If you receive an HTTP 429 response, your workspace has reached a request or token limit. Check the current Model APIs limits, then submit the event rate-limit form shared in the attendee channel.

While you wait:

• Retry 429 responses with exponential backoff.
• Avoid retrying immediately in a tight loop.
• Run independent requests concurrently, but keep the total request rate below your workspace

limit.

• Keep shared instructions and examples at the start of prompts so automatic prompt caching can

reuse them.

If the form does not resolve the issue, visit the Baseten booth or ask in #spons-baseten-2026.

Error Codes

| Code | Meaning | What to do | | ---: | --- | --- | | 400 | Invalid request | Check the model slug, request body, and model-specific parameters. | | 401 | Invalid or missing API key | Confirm BASETEN_API_KEY is set and sent as a bearer token. | | 402 | Payment required | Redeem the event credits or check your workspace billing status. | | 404 | Model not found | Check the model slug against the live model catalog. | | 429 | Rate limit exceeded | Retry with exponential backoff and use the event rate-limit form if needed. | | 500 | Internal server error | Retry the request. Ask for help if the error continues. |

See the inference troubleshooting guide for more debugging steps.

Training

We are offering H100 workstations for hackers who want to train or fine-tune their own models. Visit the Baseten booth first so the team can enable training access and help you choose an appropriate setup.

Once your workspace has access, the usual workflow is:

1. Install the Baseten CLI and sign in. 2. Define your training image, GPU requirements, commands, secrets, cache, and checkpoint storage. 3. Submit the job and follow its logs and metrics from the CLI or dashboard. 4. Save outputs under the Baseten checkpoint directory so they persist after the job stops. 5. Deploy a completed checkpoint when you are ready to test inference.

Start here:

• Training overview
• Train and deploy your first model
• Create an interactive GPU workstation
• [Connect to a training...