OpenAI's GPT-6.1 Sol can reportedly deliver performance comparable to the larger GPT-6 Astra at roughly one-fifth of the cost per task, a gap that widens as the price of Nvidia hardware rises.
AI Models · OpenAI
A smaller OpenAI model reportedly matches its larger sibling at about one-fifth the cost per task
GPT-6.1 Sol is reported to perform comparably to the larger GPT-6 Astra, and the cost gap widens as Nvidia hardware gets more expensive.
Key figures
~1/5
Cost per task for GPT-6.1 Sol versus GPT-6 Astra (reported)
Comparable
Performance level of Sol relative to the larger Astra (reported)
Rising
Nvidia hardware prices, which widen the cost gap over time
Relative cost per task
Each block is one unit of cost. Sol is shown at roughly one-fifth of Astra's per-task cost.
Takeaway: at comparable reported performance, Sol's per-task cost is about one-fifth of Astra's.
Read with caution
The one-fifth figure is a headline comparison, not a price guarantee
Workload dependent
Per-task cost varies with prompt length, output size, and how the work is structured.
Benchmarks not detailed
The specific tasks and detailed benchmark results have not been published in the available material.
Not independently verified
Third-party evaluations of the cost figures have not been confirmed.
Why it could matter
Enterprises limited by inference costs could expand AI deployments.
Smaller developers could access capabilities once reserved for the largest models.
Efficiency could offset rising compute prices for businesses running AI at scale.
What remains open
OpenAI has not explained how the efficiency gain was achieved.
Whether the parity holds for complex reasoning, coding, and multimodal tasks.
Independent benchmarking and developer reaction have not yet been verified.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article...