GPT-6.1 Sol launch: API pricing and capabilities

OpenAI announced GPT-6.1 Sol on September 29, 2026. OpenAI presents it as an upgrade to GPT-6 Sol for coding, computer use, and professional workflows. The official model ID is gpt-6.1-sol. TTAPI lists the model with its standard 20% OpenAI-model discount. OpenAI offers it in its API and in supported Work and Codex plans; its launch post says it is not yet in Chat.

OpenAI describes the model as approaching GPT-6 Astra on several tasks at one-fifth of Astra's standard input and output token rates. That is a provider comparison under OpenAI's evaluation setups, not a guarantee for every application. Start with a task your team can score, then measure the completed result, latency, and tokens used.

OpenAI reports that GPT-6.1 Sol exceeds GPT-6 Sol's best DeepSWE v1.1 score by 6.4 percentage points at lower reasoning effort and cost, and approaches Astra on that software engineering evaluation. For document-heavy professional work, its launch cites strong GDP.pdf results and an AutomationBench score 4.8 points above GPT-6 Sol at the same medium setting.

On the OSWorld 2.0 offline computer-use evaluation, OpenAI reports a seven-point gain over GPT-6 Sol at maximum reasoning effort. The test uses an agent and application environment. A plain chat-completions request does not gain that environment or its tools by changing a model string, so reproduce the result only in a comparable setup.

The release also reports fewer factual errors on difficult, user-flagged prompts at low reasoning effort: 7.7% of answers contained an error, down from 11.4% for GPT-6 Sol. OpenAI says these prompts were deliberately selected to elicit mistakes and do not represent typical usage.

OpenAI documents a 1,050,000-token context window, up to 128,000 output tokens, text and image input, and text output. The documented reasoning efforts are low, medium, high, xhigh, and max; medium is the default. The Responses API supports tool calling, while Chat Completions is documented for requests without tool calling.

For standard OpenAI API requests with no more than 272K input tokens, GPT-6.1 Sol lists $2.00 per million input tokens, $0.10 per million cached input tokens, and $10.00 per million output tokens. Applying TTAPI's existing 20% discount gives $1.6 input, $0.08 cached input, and $8 output per million tokens. Long-context requests and other processing tiers have different official rates.

OpenAI's system card addendum places GPT-6.1 Sol in its Critical cybersecurity and High biological and chemical capability categories, with the same safeguard stack as Astra. The launch post reports improvements in transparency and in respecting user restrictions on challenging evaluations. These are provider assessments, not a substitute for your own application permissions, review steps, and logging.

OpenAI's Work and Codex rollout, its Responses API tools, and any future speed modes are separate product surfaces. TTAPI's currently documented public path is a generic OpenAI-compatible chat-completions endpoint. Do not assume that every OpenAI-side tool or speed mode is available through the gateway.

The cURL below uses TTAPI's existing chat-completions request shape with the gpt-6.1-sol model ID. Check that the model is enabled for your account before routing production traffic. The example shows request syntax; it is not a recorded gateway result.

Run the same review prompt against GPT-6.1 Sol and a model you already use. Record the response, token usage, elapsed time, and whether the proposed checks actually caught the risks. Compare cost per accepted answer rather than price per token alone.

curl --request POST \
  --url 'https://api.ttapi.io/v1/chat/completions' \
  --header "Authorization: Bearer $TTAPI_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
  "model": "gpt-6.1-sol",
  "messages": [
    {
      "role": "user",
      "content": "Review this migration plan. Identify the three highest risks and a concrete acceptance check for each."
    }
  ],
  "stream": false
}'

• Confirm gpt-6.1-sol appears in TTAPI's supported models for your account.
• Check current TTAPI billing, including cached and long-context tokens.
• Keep TTAPI_KEY on the server and send no tools through this chat-completions example.
• Test quality, latency, and total cost on your own accepted tasks.

GPT-6.1 Sol and GPT-6 Sol have the same published standard input and output prices, the same 1,050,000-token context window, and the same 128,000-token output limit. Their cached-input rates differ: OpenAI lists $0.10 per million for GPT-6.1 Sol and $0.20 for GPT-6 Sol. On TTAPI's 20%-off price table, those cached-input rates are $0.080 and $0.160 respectively.

OpenAI's launch evaluations favor GPT-6.1 Sol for demanding software engineering and computer-use tasks, but a migration decision still needs your own acceptance tests. Replay a fixed set of coding reviews, document questions, and error-prone prompts through both model IDs. Score completed tasks, factual corrections, latency, and total token usage; a lower cached-input rate matters most when your requests actually reuse a substantial prompt prefix.

Check API behavior before replacing a model ID in an existing tool workflow. OpenAI documents GPT-6.1 Sol tool calling through the Responses API and Chat Completions without tool calling. Its reasoning efforts start at low, so a client that depends on GPT-6 Sol's none setting or Chat Completions function calls may need a request-flow change. TTAPI's gateway feature support should be checked separately from OpenAI's direct API features.

The GPT-6.1 Sol context window is 1,050,000 tokens, with a maximum of 128,000 output tokens. That capacity is useful for large repositories and long documents, but it is not a flat-price allowance. OpenAI states that requests above 272K input tokens use higher input, cached-input, and output rates for the entire request. Estimate cost using the actual prompt length rather than the headline context limit.

Prompt caching can reduce the cost of repeated, unchanged input. Place stable instructions and reference material before request-specific content, then check the API's usage data to see whether input was cached. OpenAI lists $0.10 per million cached-input tokens and $2.50 per million cache-write tokens for GPT-6.1 Sol at standard short-context rates. The TTAPI price table shows its discounted cached-input rate; confirm any cache-write and long-context billing in your account before forecasting a production budget.

For a fair GPT-6.1 Sol API price comparison, separate uncached input, cached input, cache writes, and output tokens. A workload with a new prompt every time will not gain the same cache savings as a repeated system prompt or shared document prefix. Record these counts alongside quality and latency when comparing GPT-6.1 Sol with GPT-6 Sol.

OpenAI's current ChatGPT pricing lists Plus at $20 per month and Pro options at $100, $200, or $500 per month. It also lists Free at $0, Go at $8, Business from $20 per user per month with annual billing, and Enterprise and Edu through sales. These are subscription prices for ChatGPT Work and Codex access, not token prices for the OpenAI API or TTAPI.

GPT-6.1 Sol is rolling out to Plus, Pro, Business, Enterprise, and Edu in Codex and ChatGPT Work. OpenAI says Free and Go are not included at launch; Enterprise and Edu administrators must enable it. The model is available in Work and Codex, not the regular Chat surface. Access still depends on the account, client, and rollout state.

Plus and Pro users can purchase additional ChatGPT credits after using their included allowance. That credit option does not change TTAPI's API billing, and a ChatGPT subscription does not supply TTAPI token balance. For a product integration, compare API token rates and gateway support; for personal Work or Codex use, compare subscription limits and any extra-credit spending.