GPT-6 Sol and Luna are now on Token Station

GPT-6 Sol and GPT-6 Luna are now available on Token Station as openai/gpt-6-sol and openai/gpt-6-luna, through the same OpenAI-compatible endpoint already serving GPT-6 Astra and the GPT-5.6 family.

Astra, OpenAI’s flagship GPT-6 release, shipped first and priced itself for the hardest agentic and computer-use work. Sol and Luna arrived three weeks later, carrying much of Astra’s underlying work into cheaper, faster tiers: Sol as a balanced route for interactive and agentic coding, Luna as the lowest-cost option in the family for high-volume, lighter-weight tasks.

What’s new

Measured against their GPT-5.6 namesakes, Sol’s list price drops 60-67% and Luna’s drops 90% or more (see Pricing below). OpenAI’s own benchmark releases position the gain as cost efficiency more than a flat capability jump:

• AutomationBench 1.0.6 (business workflows across 47 tools spanning sales, marketing, operations, support, finance, and HR): Sol at xhigh effort scores 33.2% at $0.27 per task, against Claude Opus 5 at max effort scoring 26.9% at roughly 11 times the cost.
• Agents’ Last Exam: Sol at max effort reaches 56.4%, edging past Claude Opus 5’s best recorded score at about 60% lower cost per task.
• DeepSWE v1.1 (software-engineering tasks in real codebases): Sol at max effort scores 68.8%, within 1.1 points of Claude Fable 5’s 69.9% at roughly 80% lower cost per task. Luna at max effort reaches 66.6% on the same benchmark, comparable to Claude Opus 5 and Claude Fable 5 at medium effort, at 93% and 96% lower cost per task respectively.

None of this makes Sol or Luna outright stronger than GPT-5.6 Sol or Astra on every axis. The pitch is cost: comparable or close results at a fraction of the price, which matters more for high-volume agentic fan-out than for a single hard call.

Specs

Luna’s knowledge cutoff lands later than Sol’s and even Astra’s (April 30, 2026). It’s unusual for the cheapest tier in a family to carry the most current knowledge, but that’s what OpenAI shipped here.

Try it

curl https://models.bytefuture.ai/v1/chat/completions \
  -H "Authorization: Bearer TOKEN_STATION_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-6-sol",
    "messages": [
      {"role": "user", "content": "Plan a safe refactor for a pricing module and list the tests to run."}
    ]
  }'

Swap openai/gpt-6-sol for openai/gpt-6-luna to move the same call to the cheaper tier, or for openai/gpt-6-astra to see whether the flagship route earns its higher price on your own task.

Pricing

Token Station passes these rates through directly, no markup, metered per request.

Where Sol and Luna fit

• GPT-6 Sol: interactive and agentic coding that benefits from careful, multistep validation, the middle tier between Luna’s speed and Astra’s long-horizon agentic strength.
• GPT-6 Luna: high-volume, lower-stakes work: triage, classification, exploration passes, and subtask fan-out in a larger agent, where the DeepSWE results above show it closing most of the gap to pricier models at a fraction of the cost.
• GPT-6 Astra: still the route to reach for long agentic coding sessions, computer use, and terminal-heavy operations work, where its lead over the GPT-5.6 family was largest.

A practical routing pattern: default to Luna for exploration and fan-out, move to Sol for the actual implementation and validation work, and reserve Astra for sessions that need to run the longest without drifting off task.

Get started

Sign up at models.bytefuture.ai: no card required, with a 100% match up to $50 on your first top-up. Export your key and point your existing OpenAI-compatible integration at openai/gpt-6-sol or openai/gpt-6-luna.

Try Token Station