OpenAI model
GPT-6.1 Sol
A high-performance GPT-6.1 model that brings near-Astra capability to complex coding, computer use, and professional work at a substantially lower inference cost.
- Context window
- 1.05M
- tokens
- Max output
- 128K
- tokens
- Input
- $2.00
- per 1M tokens
- Cached input
- $0.10
- per 1M tokens
- Output
- $10.00
- per 1M tokens
01 / Overview
What GPT-6.1 Sol Is
GPT-6.1 Sol is OpenAI's optimized high-performance model for complex work, positioned close to GPT-6 Astra in capability while offering a significantly lower token price.
A Practical High-Capability Production Tier
OpenAI describes GPT-6.1 Sol as delivering near-Astra performance for complex coding, computer use, and professional work. That makes it different from a purely efficiency-focused model: Sol is intended for projects where task complexity remains high, but cost and latency still need to be managed.
This positioning is useful for applications that cannot justify Astra for every difficult request but still need stronger performance than a lightweight model provides. A project may involve code, long documents, multiple constraints, tool calls, iterative refinement, or a polished deliverable that must remain internally consistent.
GPT-6.1 Sol was released as part of the GPT-6.1 family and is designed to sit in the middle of a practical production strategy: use Luna for focused frequent operations, Sol for complex work where economics matter, and Astra when maximum capability is the dominant requirement.
- Near-Astra performance for complex work at a lower cost.
- Designed for coding, computer use, and professional workflows.
- Supports a 1.05M-token context window.
- Offers five reasoning-effort levels from
lowthroughmax. - Supports modern Responses API tools and multi-agent workflows.
- Provider
- OpenAI
- Family
- GPT-6.1
- Positioning
- Near-Astra performance
- Knowledge cutoff
- Apr 30, 2026
- Input modalities
- Text, Image
- Output modality
- Text
- Default reasoning
- Medium
02 / Use cases
Where GPT-6.1 Sol Fits Best
GPT-6.1 Sol is a strong fit for complex projects where the model needs substantial reasoning and tool capability, but the workflow also runs often enough for cost and latency to remain important engineering constraints.
Complex Enough to Need Quality, Frequent Enough to Need Efficiency
Many production workloads fall between two obvious extremes. They are too difficult for a purely throughput-oriented model, yet frequent enough that using the most expensive model for every request would create unnecessary inference spend.
GPT-6.1 Sol targets that middle ground. OpenAI specifically recommends it for complex technical work and coordinated deliverables, and highlights scenarios such as building a website from a product brief or creating a board presentation from financial results.
For software teams, this can include substantial implementation tasks, debugging across multiple files, planning coordinated changes, or agentic execution that must use tools and revise intermediate work. For professional users, it can include research synthesis, polished documents, analysis based on conflicting evidence, or deliverables that combine several sources and formats.
- Complex coding and software-engineering projects.
- Coordinated deliverables that require several stages of work.
- Computer-use tasks with multiple interactions and state changes.
- Professional analysis where quality matters but Astra-level pricing is difficult to justify.
- Agentic workflows that combine tools, context, and iterative execution.
- 01
Complex technical work
Implement, debug, or coordinate software changes that require understanding multiple files and constraints.
- 02
Coordinated deliverables
Create substantial outputs such as presentations, websites, reports, or plans from complex source material.
- 03
Computer-use workflows
Work through graphical interfaces and multi-step application tasks using supported computer-use tooling.
- 04
Agentic execution
Combine tool calls, intermediate results, and delegated work in longer-running Responses API workflows.
03 / Pricing
GPT-6.1 Sol Pricing
GPT-6.1 Sol uses a substantially lower price point than Astra: $2.00 per million input tokens, $0.10 per million cached input tokens, $2.50 per million cache-write tokens, and $10.00 per million output tokens.
Cost Is Part of Sol's Product Positioning
Sol's economics are not incidental. The model is explicitly designed to deliver strong performance on complex work while reducing the cost barrier associated with a capability-first model.
Cached input is priced at five percent of the uncached input-token rate, which can make repeated stable context especially relevant for production systems with large system prompts, shared instructions, or reusable background information.
A useful cost analysis should include not only input and output tokens but also cache writes, long-context multipliers, tool usage, and processing tier. OpenAI lists Fast mode at 2× Standard pricing, while Batch and Flex are 50% lower than Standard. Regional processing can add a 10% premium where available.
- $2.00 per 1M standard input tokens.
- $0.10 per 1M cached input tokens.
- $2.50 per 1M cache-write tokens.
- $10.00 per 1M output tokens.
- Tool-specific operations may add separate charges.
1M tokens · USD
- Input
- $2.00
- Cached input
- $0.10
- Cache writes
- $2.50
- Output
- $10.00
Example: 10K input + 2K output
- Input cost
- $0.0200
- Output cost
- $0.0200
- Estimated total
- $0.0400
04 / Context
A 1.05M-Token Context Window for Complex Projects
GPT-6.1 Sol supports a 1,050,000-token context window and up to 128,000 output tokens, giving complex projects room for large source sets, long instructions, code, documents, and tool results.
Context Capacity Supports Coordinated Work
Large-context capability is particularly useful when a project depends on relationships between many pieces of information.
A software task may require repository files, implementation requirements, test failures, logs, and previous changes. A professional deliverable may combine source documents, financial data, design requirements, review comments, and prior drafts. A computer-use workflow may accumulate state and results over several interactions.
Sol's context capacity allows these inputs to remain available in a larger working set, but efficient context selection is still essential. Sending unrelated information can increase both token cost and reasoning complexity without improving the answer.
The 272K-token pricing threshold is also important when designing long-context applications. The model can technically accept far more context, but the economic cost of crossing that threshold should be measured before making very large prompts routine.
- 1,050,000-token context window.
- Maximum output of 128,000 tokens.
- Useful for codebase-level, document-heavy, and coordinated project workflows.
- Large working sets should still be filtered for relevance.
- Requests above 272K input tokens use higher pricing multipliers.
Context window
1,050,000
Max output
128,000
Context can include system instructions, project files, source documents, conversation history, retrieved evidence, tool results, and the current task.
05 / Reasoning
Reasoning Controls for Complex Work
GPT-6.1 Sol supports low, medium, high, xhigh, and max reasoning effort, with medium as the default. The none and minimal settings are not supported.
Treat Reasoning Effort as a Quality-Cost Control
Reasoning level changes how the model approaches a difficult task, so it should be evaluated as part of the model configuration rather than chosen once for the entire application.
OpenAI's model-selection guidance places GPT-6.1 Sol at medium reasoning for complex technical work and coordinated deliverables that are expected to be revised. Higher reasoning levels are positioned for polished deliverables, connected visual systems, and decisions based on conflicting evidence.
That gives teams a useful testing strategy. Start at medium for representative complex work, then increase effort for the cases where additional depth materially improves the result. The goal is to identify the lightest reasoning setting that consistently meets the product's quality bar.
Unlike some earlier Sol models, GPT-6.1 Sol does not support none or minimal reasoning effort.
Complex project baseline
Use medium reasoning as a practical starting point for substantial technical work and coordinated deliverables.
High-polish or conflicting evidence
Test higher reasoning for difficult decisions, polished outputs, and tasks that require reconciling competing information.
06 / Capabilities
Tooling for Coding, Computer Use, and Agentic Workflows
GPT-6.1 Sol supports text and image input, streaming, function calling, structured outputs, and a broad set of Responses API tools for building complex application workflows.
Use the Responses API for Tool-Driven Work
OpenAI specifically directs developers to the Responses API for tool calling with GPT-6.1 Sol. Chat Completions is supported, but without tool calling for this model.
Through the Responses API, Sol supports web search, file search, image generation, code interpreter, hosted shell, Apply Patch, Skills, computer use, MCP, and Tool Search. This gives applications a large set of building blocks for moving from reasoning to execution.
GPT-6.1 Sol also supports Multi-agent in beta. This allows a Responses API workflow to delegate work to subagents, which is relevant for larger tasks that can benefit from decomposition across specialized subtasks.
The model supports text and image input and text output. Direct audio and video modalities are not supported. Fine-tuning is currently listed as unsupported.
- Text input and output with image input.
- Streaming and structured outputs.
- Function calling through the Responses API.
- Web search and file search.
- Image generation and code interpreter.
- Hosted shell and Apply Patch.
- Skills, computer use, MCP, and Tool Search.
- Multi-agent support in beta.
- Fine-tuning is currently not supported.
- Supported
Text and vision
Accept text and image input and generate text output.
- Supported
Structured workflows
Use function calling and structured outputs through the Responses API.
- Supported
Coding tools
Use code interpreter, hosted shell, and Apply Patch for technical workflows.
- Supported
Computer use
Build workflows that interact with graphical interfaces and external applications.
- Supported
Search and integrations
Use web search, file search, MCP, Skills, Tool Search, and image-generation tooling.
- Supported
Multi-agent
Delegate work to subagents in Responses API workflows using the current beta capability.
07 / Evaluation
Strengths and Limitations
GPT-6.1 Sol is compelling when an application needs near-Astra capability but still has to optimize the cost and latency of running complex work at production scale.
Strengths
Near-Astra performance
OpenAI positions GPT-6.1 Sol close to Astra for complex work while offering substantially lower token pricing.
Strong production economics
A $2 input and $10 output rate makes high-capability workflows easier to scale than an Astra-first architecture.
Complex project fit
Coding, professional work, computer use, coordinated deliverables, and agentic execution align directly with the model's intended role.
Broad execution tooling
Responses API support includes search, coding tools, computer use, MCP, Tool Search, Skills, and multi-agent beta.
What to consider
Not the maximum-capability tier
OpenAI still positions Astra as the state-of-the-art choice when quality matters more than usage cost or latency.
No reasoning-off mode
GPT-6.1 Sol does not support none or minimal reasoning effort, which makes it less suitable for workflows that need a true zero-reasoning configuration.
Tool calling requires Responses API
Chat Completions is supported, but OpenAI directs tool-calling workloads to the Responses API.
Long-context pricing increases
Requests above 272K input tokens use higher input, cache, and output pricing multipliers for the entire request.
Evaluate the quality-cost tradeoff
Test GPT-6.1 Sol on your real complex workloads
Run the same coding, professional, or tool-driven tasks you expect in production and measure response quality, token usage, request cost, and context consumption before choosing a model tier.
Start FreeConnect your own OpenAI API key and compare GPT-6.1 Sol under the same prompt, context, and reasoning settings your application will use.
Common Questions
What is GPT-6.1 Sol?
GPT-6.1 Sol is an OpenAI model designed to deliver near-Astra performance at a lower cost for complex coding, computer use, and professional work.
How much does GPT-6.1 Sol cost?
Standard pricing is $2.00 per 1M input tokens, $0.10 per 1M cached input tokens, $2.50 per 1M cache-write tokens, and $10.00 per 1M output tokens. Tool-specific charges may apply separately.
What is the context window of GPT-6.1 Sol?
GPT-6.1 Sol has a 1,050,000-token context window and supports up to 128,000 output tokens.
What reasoning levels does GPT-6.1 Sol support?
The model supports low, medium, high, xhigh, and max reasoning effort. Medium is the default. None and minimal reasoning are not supported.
What is the knowledge cutoff for GPT-6.1 Sol?
OpenAI lists April 30, 2026 as the knowledge cutoff for GPT-6.1 Sol.
Does GPT-6.1 Sol support image input?
Yes. GPT-6.1 Sol supports text and image input and produces text output. Direct audio and video modalities are not supported.
Does GPT-6.1 Sol support tool calling?
Yes. OpenAI recommends the Responses API for tool calling with GPT-6.1 Sol. Chat Completions is supported without tool calling.
What tools does GPT-6.1 Sol support?
The current model documentation lists web search, file search, image generation, code interpreter, hosted shell, Apply Patch, Skills, computer use, MCP, and Tool Search as supported Responses API tools.
Does GPT-6.1 Sol support multi-agent workflows?
Yes. OpenAI introduced Multi-agent support in beta for GPT-6.1 Sol, allowing a Responses API request to delegate work to subagents.
When should I choose GPT-6.1 Sol instead of Astra?
GPT-6.1 Sol is intended for complex projects where strong model capability is required but cost and latency still matter. Astra remains the maximum-capability option when quality is the overriding priority.
Can GPT-6.1 Sol be fine-tuned?
No. The current OpenAI model documentation lists fine-tuning as unsupported for GPT-6.1 Sol.
Model information
Last updated
The specifications and prices on this page are based on the official OpenAI documentation for GPT-6.1 Sol. Provider pricing, tools, processing tiers, residency options, limits, and availability may change, so production assumptions should be checked against the latest provider documentation.