Updated: October 4, 2026. OpenAI’s GPT-6.1 Sol is positioned as a high-capability model for coding, computer use and professional work without the price of its flagship GPT-6 Astra model.

The release matters because the gap between “best model” and “best model for the money” is becoming more important for developers and businesses. A model that performs close to the flagship on real workflows can be more valuable if it costs far less to run at scale.

GPT-6.1 Sol at a glance

  • Model: GPT-6.1 Sol
  • Release date: September 29, 2026
  • Standard API input: $2 per 1 million tokens for prompts within the standard pricing band
  • Cached input: $0.10 per 1 million tokens
  • Output: $10 per 1 million tokens
  • Context window: 1,050,000 tokens
  • Maximum output: 128,000 tokens
  • Reasoning effort: low, medium, high, xhigh and max
  • Primary positioning: near-Astra performance for complex work at lower cost

How much does GPT-6.1 Sol cost?

OpenAI lists standard pricing of $2 per million input tokens, $0.10 per million cached input tokens and $10 per million output tokens for requests within the standard input band.

For comparison, GPT-6 Astra is priced substantially higher. That means teams running large volumes of coding, research or agentic workloads may see a major cost difference even when the quality gap is small for their specific task.

There are additional pricing rules for very long context and different service tiers, so developers should check the current model documentation rather than assuming one flat price applies to every request.

Why is the cached-input price important?

Many agent systems repeatedly send the same system prompt, tool descriptions, policies or project context. If those repeated tokens can be served from cache, the effective cost of a long-running workflow can fall sharply.

That makes GPT-6.1 Sol especially interesting for applications with stable context reused across many calls, such as coding agents, document-analysis systems and multi-step business workflows.

How does GPT-6.1 Sol compare with GPT-6 Astra?

OpenAI describes GPT-6.1 Sol as delivering capabilities comparable to Astra at a lower cost. The company’s release materials and third-party benchmark summaries show Sol close to Astra on several coding and computer-use evaluations, although benchmark results should not be treated as a guarantee for every workload.

The practical comparison is straightforward:

  • Astra: choose it when maximum capability is worth premium cost.
  • GPT-6.1 Sol: choose it when you want strong reasoning and agentic performance but need much better cost efficiency.
  • Cheaper/faster models: use them for high-volume tasks where the extra reasoning ability is unnecessary.

What kinds of work is GPT-6.1 Sol designed for?

Coding and software engineering

OpenAI highlights complex coding as one of the model’s main strengths. That includes tasks requiring multiple files, tool use, debugging, implementation planning and sustained reasoning rather than only short code completion.

Computer use and agentic workflows

GPT-6.1 Sol is also aimed at workflows where the model uses tools and acts across software environments. In the API, OpenAI recommends the Responses API for tool calling.

Professional knowledge work

The large context window makes the model suitable for lengthy documents, research collections, financial models, specifications and other work where the system must keep track of a large amount of information.

Does GPT-6.1 Sol support tool calling?

Yes, but endpoint choice matters. OpenAI’s current documentation says to use the Responses API for tool calling. Chat Completions is supported without tool calling.

This is an important implementation detail for developers migrating older applications. A model upgrade may require an API architecture change if the application depends heavily on tools.

What should businesses test before switching?

Do not switch based only on a benchmark table. Run a small evaluation set that represents your actual work.

  1. Choose 20–50 real tasks from production.
  2. Score correctness, completeness and formatting.
  3. Track average token use and full task cost.
  4. Measure latency and tool-call reliability.
  5. Compare how often a human needs to fix the output.

A cheaper model that requires significantly more human correction may not actually be cheaper. Conversely, a model that is only slightly below the flagship on a benchmark may be the better business choice if it completes your real tasks reliably at a fraction of the cost.

What does this release say about the AI market?

The rapid model cycle is shifting competition toward price-performance rather than raw capability alone. Businesses increasingly have access to multiple model tiers, and the best architecture may route easy tasks to cheaper models while reserving premium models for the hardest work.

BCC recently covered how the AI value chain is expanding across models, infrastructure and applications in our look at AI companies and the 2026 AI value chain.

Frequently asked questions

What is GPT-6.1 Sol?

It is an OpenAI reasoning model released on September 29, 2026 for complex coding, computer use and professional work.

What is GPT-6.1 Sol pricing?

OpenAI lists standard pricing of $2 per million input tokens, $0.10 per million cached input tokens and $10 per million output tokens within the standard input band.

What is the GPT-6.1 Sol context window?

The model documentation lists a 1,050,000-token context window and up to 128,000 output tokens.

Is GPT-6.1 Sol better than GPT-6 Astra?

Astra remains the premium model. GPT-6.1 Sol is designed to get close to Astra on many complex tasks at a much lower cost. The better choice depends on the workload.

Sources and further reading

Follow Buzz Content Corner

Want more timely explainers and verified updates? Follow Buzz Content Corner (BCC) on Instagram and Facebook.