Short answer: GPT-6.1 Sol and GPT-6.1 Terra are the newest releases in OpenAI's GPT-6 family. GPT-6.1 Sol (released September 29, 2026) offers what OpenAI calls "near-Astra performance for complex work at a lower cost," at $2 per million input tokens and $10 per million output tokens. GPT-6.1 Terra (released October 26, 2026) costs half that, $1 and $5, and OpenAI positions it for visual reasoning, precise computer use, and clear, natural writing. Both have a 1,050,000-token context window and up to 128,000 output tokens. If your team wants either model inside Slack or Microsoft Teams, you can choose it in a Runbear agent's model settings.
Disclosure: Runbear is our product. OpenAI facts and prices come from OpenAI's developer documentation as of October 2026 and are linked inline. Check those pages before you plan a budget.
What OpenAI released
GPT-6.1 Sol and Terra are the newest releases in the GPT-6 family. OpenAI's API changelog shows the order:
- September 3, 2026: GPT-6 Astra, described as OpenAI's "most capable model, built for the hardest end-to-end work."
- September 22, 2026: GPT-6 Sol and GPT-6 Luna.
- September 29, 2026: GPT-6.1 Sol, "for complex coding and professional work at a lower cost than GPT-6 Astra."
- October 26, 2026: GPT-6.1 Terra, "a fast, efficient model for visual reasoning, precise computer use, and clear, natural writing."
The family uses tier names (Astra, Sol, Terra, Luna) rather than size suffixes like mini or nano. A tier can move to a new release on its own schedule, which is why the current lineup mixes GPT-6 (Astra, Luna) and GPT-6.1 (Sol, Terra). In its October 7 changelog entry, OpenAI wrote: "We recommend leveraging the GPT-6 model family for production API usage."
GPT-6.1 Sol vs GPT-6.1 Terra at a glance
Every row below comes from OpenAI's model pages for GPT-6.1 Sol and GPT-6.1 Terra, as of October 2026.
| GPT-6.1 Sol | GPT-6.1 Terra | |
| OpenAI's positioning | "Near-Astra performance for complex work at a lower cost" | "A fast, efficient model for visual reasoning, precise computer use, and clear, natural writing" |
| Input (per 1M tokens) | $2.00 | $1.00 |
| Cached input | $0.10 | $0.05 |
| Cache writes | $2.50 | $1.25 (30-minute retention) or $2.50 (12-hour retention) |
| Output (per 1M tokens) | $10.00 | $5.00 |
| Context window | 1,050,000 tokens (922,000 max input) | 1,050,000 tokens (922,000 max input) |
| Max output | 128,000 tokens | 128,000 tokens |
| Knowledge cutoff | April 30, 2026 | June 1, 2026 |
| Reasoning effort | low, medium (default), high, xhigh, max | none, low, medium (default), high, xhigh, max |
| Input and output | Text and image in, text out | Text and image in, text out |
Both models list the same Responses API tools, including web search, file search, code interpreter, computer use, and MCP. OpenAI notes that tool calling requires the Responses API. Both also share a long-context rule: prompts with more than 272K input tokens are billed at 2x the input and cache rates and 1.5x the output rate for the whole request.
What changed from GPT-6 Sol
GPT-6.1 Sol is the newer Sol release; GPT-6 Sol still has its own model page. The headline prices are the same, but three details differ:
- Cheaper cached input. Cached input drops from $0.20 to $0.10 per million tokens, or 5% of the uncached rate instead of a tenth. Agents that reuse a long system prompt or the same documents on every turn benefit the most.
- No "none" reasoning effort. GPT-6 Sol accepts
nonethroughmax. The GPT-6.1 Sol page listslowthroughmax. If you relied onnonefor the fastest replies, GPT-6.1 Terra still supports it. - A newer knowledge cutoff. April 30, 2026, up from April 20, 2026.
OpenAI also added an Ultrafast service tier for GPT-6.1 Sol on October 8 (service_tier: "ultrafast"). The model page prices Ultrafast at 6x Standard.
What one request costs
At list API prices, a request with 20,000 input tokens and 1,000 output tokens, none of them cached, costs:
- GPT-6.1 Sol: 20,000 × $2/1M + 1,000 × $10/1M = $0.05
- GPT-6.1 Terra: 20,000 × $1/1M + 1,000 × $5/1M = $0.025
The long-context rule changes the math quickly. A Sol request with 300,000 input tokens crosses the 272K line, so the whole request is billed at $4 input and $15 output per million: 300,000 × $4/1M + 1,000 × $15/1M = $1.215, about twice the $0.61 it would cost at the base rate. If an agent searches a knowledge base and sends only the relevant passages, it stays far below that line.
These are OpenAI API prices. Inside Runbear, an agent on Runbear-provided access uses Runbear credits instead, as described below.
Which one should your team pick
OpenAI's own descriptions are the best starting point:
- GPT-6.1 Sol for multistep work: complex coding, computer use, and professional analysis where the model has to plan and call several tools in a row.
- GPT-6.1 Terra for writing-heavy work and visual reasoning at half the price: drafting replies, summaries, and status updates, or reading screenshots and technical drawings.
- A lite tier for lookups. Runbear's agent settings docs note that if an agent mostly answers from a knowledge base, the cheapest tier usually matches the most expensive one on answer quality while consuming a fraction of the credits.
A more capable model is not automatically a better agent. Start with the cheaper tier that fits the job, run the questions your team actually asks, and move up only where the answers fall short.
How to use GPT-6.1 Sol or Terra in Slack with your team
When a team wants one agent that everyone in a channel can @mention, with shared instructions, knowledge, and tools, the agent needs to live in Slack. Runbear is a no-code AI agent builder for Slack that lets you pick the model per agent, including GPT-6.1 Sol and Terra. For a fuller comparison with OpenAI's own Slack options, see Runbear vs ChatGPT.
Setting it up takes four steps:
- Open the agent in Runbear and go to the Settings tab. The Model Selection section on Claude Agent SDK agents offers Claude, GPT, and Gemini models.
- Choose GPT Sol or GPT Terra. These rows carry a Stable badge and today follow GPT-6.1 for both tiers. To keep the agent on one exact release, choose it under Specific version instead.
- Connect the agent to Slack from the Chatbots tab and invite the bot to the channels where it should answer. The Slack setup guide walks through the options.
- Ask it something. The agent always answers @mentions and direct messages, and replying without an @mention is a separate switch you can turn on per channel.
The same agent can also answer in Microsoft Teams, since Claude Agent SDK agents support both Slack and Teams. Agents on the OpenAI Responses runtime offer GPT-6.1 Sol and GPT-6.1 Terra as well, but that runtime has no Stable rows, so an agent stays on the model it was saved with until someone picks another.
On Claude Agent SDK agents, Runbear-provided access uses Runbear credits, and plans are priced on usage credits rather than seats. See pricing for the current plans.
FAQ
When was GPT-6.1 Sol released?
OpenAI released GPT-6.1 Sol on September 29, 2026, according to its API changelog.
When was GPT-6.1 Terra released?
October 26, 2026. OpenAI describes it as "a fast, efficient model for visual reasoning, precise computer use, and clear, natural writing."
How much does GPT-6.1 Terra cost?
$1 per million input tokens, $0.05 per million cached input tokens, and $5 per million output tokens on the Standard tier, as of October 2026. Prompts over 272K input tokens are billed at 2x input and cache rates and 1.5x output for the whole request.
What is the difference between GPT-6.1 Sol and GPT-6.1 Terra?
OpenAI describes Sol as "near-Astra performance for complex work at a lower cost" and Terra as a fast, efficient model for visual reasoning, precise computer use, and writing. Terra costs half as much per token, has a later knowledge cutoff (June 1 vs April 30, 2026), and supports a none reasoning effort that Sol does not. Both share the same context window, max output, and tool list.
Can I use GPT-6.1 in Slack?
Yes. Build an agent in Runbear, choose GPT Sol or GPT Terra in its model settings, and connect it to a Slack channel. Your team can then @mention it in channels or message it directly. The same agent can also answer in Microsoft Teams.
Will a Runbear agent move to new GPT releases automatically?
On Claude Agent SDK agents, yes, if you choose a Stable row such as GPT Sol or GPT Terra: it follows the latest available version in that tier. An agent set to a specific version stays on it. Agents on the OpenAI Responses runtime have no Stable rows and keep the model they were saved with.
