OpenAI · Everyday work and programming
GPT-6 Sol
A versatile choice for working with text, code and images.
- Costs
- $2 input / $10 output per million tokens. Above 272 thousand input tokens: 2× input and 1.5× output.
- Speed
- We have not independently measured speed on a comparable basis.
- Availability
- OpenAI API
- Input length
- 1,050,000 tokens in total; output capped at 128 thousand tokens.
- Inputs
- Text and images → text
Data checked . Published September 22, 2026. Specifications and prices are provided by the vendor; usage recommendations come from the editorial team. We do not yet have our own comparative test of this version.
Ideal use
When to choose it
- Writing and fixing code
- Processing documents and multi-step tasks
Usage boundaries
When to choose another model
- For simple classification, consider the cheaper GPT-6 Luna.
Traceable supporting sources
Data and measurement sources
Distinguish between the manufacturer's documentation and the results of a specific test. Measurements also depend on the settings and the task set used.
Vendor documentation
Specifications, availability, and termsManufacturer data verified as of the review date. Recommended use is an editorial interpretation.
Open original source ↗Efficiency in practice
With what GPT-6 Sol combine
Many tasks without unnecessary costs
A low-cost model handles routine work, and the more expensive one gets only the truly difficult cases.
Use your own data to verify when a stronger model should take over the work. A model cannot reliably assess its own answer.Quick code edits with a second check
One system makes the change, another looks for blind spots, and tests decide.
A second model will not help if both receive the same incorrect assumption without evidence.Documents → analysis → verification
A fast model prepares the material, and a stronger model focuses on analyzing it.
Agreement between two models is not proof; the cited inputs are what matters.Trends over time
Related events from AI Radar
OpenAI halved prices for the GPT-6 Sol and Luna models; independent analysis does not confirm a performance improvement
New: Anthropic released Claude Opus 5.5 on the same day (2026-09-23), roughly an hour after OpenAI; Claude Opus 5.5 is 20% cheaper than the previous Opus 5.0 ($4/$20 vs $5/$25); Opus 5.5 improved communication style and token efficiency; The model releases launch a 'price war' between the two largest AI companies; GPT-6 Luna and Sol are part of a broader strategy of price competition, not an isolated release
Developers working with the API from OpenAI can reduce inference costs by up to half while maintaining the same usage pattern (Sol for more complex tasks, Luna for high-volume simple tasks), but independent tests suggest that a lower price may not mean higher or even the same output quality in all cases.
Businesses gain a cheaper option for reasoning models for high-volume or repetitive tasks (summaries, data extraction, code), but according to an independent analysis, actual output quality has not improved compared with the previous generation and has instead declined on some knowledge work tasks, so deployment decisions should not rely solely on the manufacturer's claims about price/performance.
OpenAI makes GPT-6 Sol and GPT-6 Luna available on the Amazon Bedrock platform
Developers working with the Amazon Bedrock platform gain another model choice based on the task type: assign more demanding tasks to GPT-6 Sol and simple, frequent tasks to GPT-6 Luna, and adjust the reasoning effort level as needed.
Companies running AI tasks on the Amazon Bedrock platform can deploy newer models from OpenAI at a lower price than GPT-5.6, with enterprise security controls (IAM, CloudTrail, VPC PrivateLink) and a choice between a model for complex tasks and a model for high volumes of simple requests.
OpenAI fixed a bug in image encoding in the GPT-6 Sol and GPT-6 Luna models
Developers who used image inputs with the GPT-6 Sol or GPT-6 Luna models may have encountered worse results due to a bug in image encoding; after the fix, it is advisable to retest your tasks involving visual content.
Companies running pipelines with image inputs via the GPT-6 Sol or GPT-6 Luna models (API, Codex, computer use) may have received worse results on visual tasks due to a bug; it is now advisable to re-verify output quality and, if needed, recompute the affected workflows.
GPT-6.1 Sol from OpenAI is generally available on Amazon Bedrock
New: GPT-6.1 Sol is available on Amazon Bedrock; It matches the performance of GPT-6 Astra on the DeepSWE v1.1 test set; On DeepSWE v1.1, it outperforms GPT-6 Sol by 6.4 % with lower reasoning effort; Availability is expanding to AWS infrastructure through a third-party provider
Developers using Codex or the API can now get coding and document-handling performance that, according to OpenAI, approaches GPT-6 Astra at one-fifth of the price of that model, including access through Amazon Bedrock infrastructure.
Companies running AI agents on AWS gain access to a model with near-flagship quality at a lower price, with enterprise controls (IAM, CloudTrail, VPC PrivateLink, isolation without access by AWS operators) and no obligation to share data with OpenAI for training — making governance and compliance decisions about deployment easier.
Next step
Compare the model using your actual task.
A benchmark narrows the selection. A short trial on your data determines the choice.