Heabsy

Heabsy is an inference platform serving open models through an OpenAI-compatible API. The LLMTR catalog carries two models plus a time-limited free row for one of them; all are addressed with the heabsy/ prefix, exactly as the catalog spells them.

Model	Context	Max output	Input / Cache read / Output ($/1M)	Image input	Reasons by default	EU-hosted
`heabsy/qwen3.8-27b-heabsy`	262,144	32,768	0.40 / 0.05 / 3.00	yes	yes	yes
`heabsy/qwen3.8-27b-heabsy-free`	262,144	32,768	0 / 0 / 0	yes	yes	yes
`heabsy/mimo-v2.6-flash-heabsy`	1,048,576	65,536	0.4762 / 0.19 / 1.4286	yes	yes	no

The other providers serving the same model are listed on each model page, in the menu next to the provider name. The other EU-hosted models are listed in the model catalog.

Heabsy models cannot be called until your account has granted the data-transfer consent for Heabsy; without it every request returns 403 provider_consent_required.

Consent is given, and withdrawn, from the "Data-transfer consents" card in Settings. It is account-level and applies to every API key on that account. Consent is per provider: a consent given for GreenPT or TensorX does not cover the Heabsy models.

The content of your request is transferred to Heabsy to produce the response and processed by Heabsy on your behalf; account details such as your name and e-mail are not transferred. See Heabsy's data processing agreement (DPA) for the details.

Free Qwen3.8 27B

heabsy/qwen3.8-27b-heabsy-free is free until 2 November 2026, 00:00 (TRT), with a daily usage limit per account. After that date requests return 410 model_retired naming the metered heabsy/qwen3.8-27b-heabsy; to keep using the same model, switch the model id to that metered row.

Chat

curl "$LLMTR_BASE_URL/v1/chat/completions" \
  -H "Authorization: Bearer llmtr-your_key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "heabsy/qwen3.8-27b-heabsy",
    "messages": [
      { "role": "user", "content": "Summarise this stack trace in three bullet points." }
    ],
    "max_tokens": 1024
  }'

Every model supports tool/function calling. max_tokens cannot exceed the max output in the table.

Streaming on these rows is not real streaming for now: stream: true is accepted, but the answer is delivered in chunks after it completes. On long answers, set how long your client waits for the first chunk accordingly.

Reasoning

On models that reason by default those tokens bill as output. If a response comes back with finish_reason: "length", raise max_tokens and retry.

reasoning_effort is not supported on this provider; a request that sends it is refused with 400 invalid_request.

Image input

Images can only be sent to the models marked "yes" for image input in the table. On the others a request containing an image is refused with 400 unsupported_input and is not billed.