Editorial illustration for OpenAI's GPT-6.1 Sol Matches Astra's Coding at Lower Cost
GPT-6.1 Sol Matches Astra's Coding at 80% Less Cost
OpenAI shipped GPT-6.1 Sol this week, a refresh of the mid-tier model in its GPT-6 lineup, and the pitch comes down to a single number. The company says Sol handles agentic coding, computer use, and professional document work at roughly the level of its flagship GPT-6 Astra model, for one-fifth the price. Input and output tokens both get cheaper, but the bigger move is on cached input, which falls to $0.10 per million tokens, half of what GPT-6 Sol charged. That matters more than it sounds, because agents burn through cached tokens fast, resending the same system prompts and tool schemas at every step of a task.
The model is live now, not a preview. Developers can call it through the OpenAI API as gpt-6.1-sol, and it's already built into ChatGPT Work and Codex. That puts Sol into a three-tier structure alongside Astra at the top and the cheaper GPT-6 Luna at the bottom, each priced for a different job. OpenAI backed the release with benchmark claims against its own Astra model and against Anthropic's Claude Opus 5.5, numbers worth looking at directly before deciding what "near-Astra" actually means in practice.
OpenAI’s claim is specific: near-Astra results on agentic coding, computer use, and professional work. The price is one-fifth of GPT-6 Astra’s standard input and output rates.
Why this matters
For teams picking a default coding model, price-per-quality just moved. If GPT-6.1 Sol really lands within a few points of Astra on DeepSWE v1.1 at a fifth of the cost, the calculus for anyone running agentic coding workloads at scale shifts toward Sol almost by default, and Astra becomes the model you reach for only when you need the last few percentage points. That's a real threat to the usual "pay more for the flagship" pattern, and it's worth testing against your own codebase, not just OpenAI's benchmark suite.
We'd flag the obvious caveat: these are OpenAI's numbers, from OpenAI's launch post, measured on OpenAI's chosen benchmark. Competitor scores are pulled from public reports, not independently re-run, so the "matches Astra" claim deserves a skeptical eye until third parties replicate it on Codex or ChatGPT Work workloads. Cached input at $0.10 per million tokens is the more concrete, verifiable number here, and it's the one worth watching if you're running high-repetition agentic loops. Availability today in the API and ChatGPT Work means you can check the claim yourself within the hour.
Related Reading
Browse all OpenAI news →
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Related Topic OpenAI's GPT Transcribe Cuts Error Rate to 3.31%, Improving on GPT-4o
Open Source
Related Topic OpenAI Developer: AI Agent "Swarms" Waste Tokens, Create "Coordination Tax
AI Tools & Apps
From Archives NVIDIA's SCADA Server SDK Lets Partners Build Storage for GPU Clients
From Archives Anthropic Warns Zhipu's GLM-5.3 Model Quickly Used to Build Exploits
From Archives OpenAI, Meta Muse Glasses Could Offer Fastest Way to Interact With AI
AI Tools & Apps
From Archives FTC probes OpenAI, Anthropic over consumer protection concerns
Business & Startups
Common Questions Answered
How does GPT-6.1 Sol's pricing compare to GPT-6 Astra for agentic coding tasks?
GPT-6.1 Sol is priced at approximately one-fifth the cost of GPT-6 Astra while delivering near-equivalent performance on agentic coding, computer use, and professional document work. The most significant savings come from cached input tokens, which dropped to $0.10 per million tokens—half the previous rate—making it substantially more economical for teams running large-scale coding workloads.
What specific capabilities does GPT-6.1 Sol handle at Astra-level performance?
According to OpenAI, GPT-6.1 Sol matches Astra's performance on agentic coding, computer use, and professional document work. This means teams can leverage Sol for complex coding tasks and automated computer interactions without sacrificing quality compared to the flagship model.
Why does the cached input token pricing matter more than standard token pricing for GPT-6.1 Sol?
Cached input tokens are crucial for applications that reuse large context windows, such as agentic coding systems that repeatedly reference the same codebase or documentation. By reducing cached input costs to $0.10 per million tokens, GPT-6.1 Sol significantly improves the economics of long-running, repetitive AI workflows that would otherwise accumulate substantial token expenses.
How does GPT-6.1 Sol change the decision-making process for teams choosing a default coding model?
With GPT-6.1 Sol delivering near-Astra results at one-fifth the cost, teams now have a compelling reason to make Sol their default coding model for most workloads, reserving Astra only for tasks requiring the highest performance margins. This shifts the traditional "pay more for the flagship" pricing pattern and makes cost-per-quality a primary factor in model selection for agentic coding at scale.
Further Reading
- Introducing GPT-6.1 Sol - OpenAI
- OpenAI's GPT-6.1 Sol offers Astra-like performance at 1/5th price. A new Ultrafast tier clocks at 300 tokens per second. - VentureBeat
- OpenAI Launches GPT-6.1 Sol With Lower Costs and Near-Astra Performance - iPhone in Canada
- GPT-6.1 Sol Release Guide: Near-Astra Agentic Work at $2/$10 and ... - Developers Digest
- GPT-6.1 Sol: Features, Benchmarks, Pricing, and Access - DataCamp
