GPT-6.1 Sol & Terra: Pricing & API Access

Introducing GPT-6.1 Sol and GPT-6.1 Terra

GPT-6.1 Sol is the flagship model of the GPT-6.1 generation. It targets multi-step reasoning, advanced coding, research synthesis, architecture reviews, and long-running agent workflows, and it sits above the existing GPT-6 Sol tier. The exact request model is gpt-6.1-sol.

GPT-6.1 Terra is the balanced option in the same generation. It handles everyday coding, extraction, classification, summarization, and routine agent steps at a much lower weight, making it the default choice for high-volume production traffic. Its exact request model is gpt-6.1-terra.

Both models support a 1M token context window and vision inputs, and both are available on the same OpenAI-compatible endpoint as the rest of the catalog. The earlier GPT-6 Sol, GPT-6 Luna, and GPT-5.6 family remain available, so existing integrations keep working while you adopt the 6.1 models.

Official pricing and model weightage

The catalog lists the following official input/output pricing references. APItokendeal usage is prepaid, and model weightage determines how quickly token-pack credits are deducted.

Check the live model catalog for current pricing and weightage before budgeting a production workload.

Which GPT-6.1 model should you use?

Choose GPT-6.1 Sol for coding agents, multi-file refactoring, architecture decisions, difficult debugging, research, and tasks where an incorrect answer costs more than additional tokens.

Choose GPT-6.1 Terra for customer support, summarization, extraction, classification, documentation, routine code assistance, and high-volume requests where latency and cost are the priority.

A practical routing pattern is to send every request to Terra first, then escalate only requests that need deeper planning or stronger reasoning to Sol. For the cheapest background tasks, GPT-6 Luna and GPT-5.6 Luna remain the lowest-weight options in the catalog.

How to access GPT-6.1 Sol and Terra

APItokendeal uses an OpenAI-compatible API. Copy the API key from your Account page, set the base URL, and choose either exact model ID.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.apitokendeal.com/v1",
    api_key="your_apitokendeal_key"
)

for model in ["gpt-6.1-sol", "gpt-6.1-terra"]:
    response = client.chat.completions.create(
        model=model,
        messages=[{"role": "user", "content": "Hello"}]
    )
    print(model, response.choices[0].message.content)

Keep the same endpoint and key when switching models. Only the model value changes.

What this means for API users

The GPT-6.1 update gives developers a clear upgrade path: Sol takes over frontier reasoning work from GPT-6 Sol at the same catalog reference price, while Terra fills the balanced middle tier between Luna-class efficiency and flagship capability. Both models run through the same OpenAI-compatible client and the same prepaid token packs.

As with the previous generation, model routing is usually more effective than running every request on the premium tier. It lets a single application balance response quality, latency, and token consumption according to the actual difficulty of each task.