SayGM (SN28) Lists GPT-6 Sol and Luna at 18.4% Off

OpenAI released GPT-6 Sol and GPT-6 Luna on September 22. Within hours, Bittensor’s privacy-focused AI gateway SayGM had both models available through its OpenAI-compatible API.

GPT-6 sol and GPT-6 luna are live on Gm.

Sol: complex coding and professional work at a lower price point.
Luna: fast, efficient everyday work at scale.

Both priced 12.9% below list from launch. https://t.co/P5ILFQvFt6

— Good Morning (@say_gm_) September 22, 2026

The fast listing was notable on its own, but the pricing makes it more interesting. SayGM launched the models at 12.9% below OpenAI’s list price, according to its original announcement. Since then, competition between miners on the subnet has moved the best available discount to 18% for both models.

Why These Models Matter

Sol is built and optimized for harder work like complex coding, multi-step professional tasks, and agent workflows. 

On OpenAI’s tests it scores 33.2% on AutomationBench (real business workflows) and 68.8% on DeepSWE (real software engineering problems), while making roughly half as many factual mistakes as the previous version. 

Luna is the cheaper, faster option for everyday high-volume jobs such as summarising documents, extracting information, or handling simpler coding steps. It still reaches 66.6% on DeepSWE scoring close to Sol at a much lower cost per task.

In short, teams can now run more capable models on the work they do every day without the bill rising as quickly.

Current Pricing on SayGM

OpenAI’s official list prices are $2 input / $10 output per million tokens for Sol and $0.10 / $0.50 for Luna. On SayGM the best available rates right now are $1.63 / $8.16 for Sol and $0.08 / $0.41 for Luna. 

That is an 18.4% discount to OpenAI’s list price for both models. Live numbers are here: For Sol and for Luna.

When SayGM first announced the models, the discount stood at 12.9%. Provider bidding has since improved it to the current 18.4%. 

This is due to the fact that more miners started offering the new models, or existing miners improved their bids. 

As competition increased, the best available rate improved to the current 18.4%. This is normal behaviour on the subnet. Prices are live and move with supply. The same mechanism is why many other models on SayGM also sit below official list prices.

See the full live catalogue HERE.

Check below to understand more on  how SayGM (SN28) is able to offer models below the official list prices:

How GM (SN28) Sells Fable and Astra for Less Than the Companies That Make Them

How the Private Routing Wworks

The gateway runs inside an Intel TDX trusted execution environment on Bittensor subnet 28, so the operators themselves SayGM in this case) never see your prompts in the clear.

For teams that care about both price and privacy, this is a practical same-day option instead of waiting for every other aggregator to catch up. 

Read more on SayGM →

Stripe’s $7B OpenRouter Bet Makes Bittensor’s GM Look Wildly Undervalued