Glonce

OpenAI cuts GPT-5.6 Sol pricing 20% on input, 33% on output

modelsThe Glonce editorsHacker News🔥74Glonce rating74Impact75Novelty35Relevance100Surprise50Players100Credibility90
OpenAI cuts GPT-5.6 Sol pricing 20% on input, 33% on output

OpenAI's API documentation page for GPT-5.6 Sol, the frontier model in the GPT-5.6 family, now lists a lower price: $4.00 per million input tokens and $20.00 per million output tokens. The page describes this as a 20% reduction in input pricing and a 33% reduction in output pricing. Cached input is priced separately at $0.40 per million tokens. The promotional pricing is guaranteed to hold at least through November 21, 2026; the page does not say what happens after that date.

The documentation's own quick comparison table lists GPT-5.5 at $5.00 and GPT-5.4 at $2.50, alongside GPT-5.6 Sol's $4.00, though the table does not label which price column (input, cached input, or output) that single figure per model represents.

GPT-5.6 Sol carries a 1,050,000-token context window and a 128,000-token maximum output, with a knowledge cutoff of February 16, 2026. The model accepts text and image input and produces text output only; audio and video are not supported. Its reasoning effort parameter can be set to none, low, medium (the default), high, xhigh, or max. Two pricing conditions apply beyond the headline rate: prompts exceeding 272,000 input tokens are billed at 2x the input rate and 1.5x the output rate for the entire request, and cache writes are billed at 1.25x the standard uncached input rate. The gpt-5.6 alias routes requests to this model, and it is exposed across OpenAI's standard endpoints, including Chat Completions, Responses, Realtime, Assistants, and Batch; fine-tuning is listed as a feature not supported for this model. Supported tools include web search, file search, image generation, code interpreter, hosted shell, apply patch, skills, computer use, MCP, and tool search. No release date, rationale for the cut, or named OpenAI spokesperson accompanies the pricing update on the page.

Key facts

  • GPT-5.6 Sol now costs $4.00 per million input tokens and $20.00 per million output tokens, described as a 20% cut on input and a 33% cut on output.
  • Cached input is billed separately at $0.40 per million tokens.
  • The promotional pricing is guaranteed at least through November 21, 2026; what happens afterward is not stated.
  • Requests with more than 272,000 input tokens are billed at 2x the input rate and 1.5x the output rate for the whole request, and cache writes cost 1.25x the uncached input rate.
  • The model has a 1,050,000-token context window, a 128,000-token max output, and a February 16, 2026 knowledge cutoff.

Why it matters

A 20 to 33 percent price cut on OpenAI's frontier-tier model lowers the cost of running complex, high-reasoning workloads through the API, which is typically the most expensive tier developers use. Cutting output pricing more sharply than input pricing (33% versus 20%) specifically targets tasks that generate long responses, such as extended reasoning traces or long-form generation, where output tokens dominate the bill.

Who it affects

Developers and businesses calling the GPT-5.6 Sol model, or the gpt-5.6 alias that routes to it, through OpenAI's API endpoints (Chat Completions, Responses, Assistants, Batch, and others). Anyone budgeting API spend around this specific model tier sees an immediate rate change; the two adjacent tiers in the documentation's comparison table, GPT-5.5 and GPT-5.4, are shown alongside it but are not stated to be changing.

How to use it

The new rate applies automatically to the gpt-5.6 alias and to direct calls to the gpt-5.6-sol snapshot; no opt-in is described. Input costs $4.00 per million tokens ($0.40 for cached input) and output costs $20.00 per million tokens. Two surcharges apply outside the headline rate: prompts over 272,000 input tokens are billed at 2x input and 1.5x output for the full request, and cache writes cost 1.25x the uncached input rate. The model supports a reasoning effort parameter (none, low, medium, high, xhigh, max) that lets callers trade cost and latency against reasoning depth, and it accepts text and image input with text-only output.

How solid is it

The figures come directly from OpenAI's own API documentation page for the model, which states the percentage reductions and the guaranteed promotional window explicitly. The page does not name a spokesperson, give a release date, or state a rationale for the cut, and it does not specify what the rate does after November 21, 2026.

Risks and caveats

The pricing is explicitly promotional and guaranteed only 'at least through' November 21, 2026, meaning it could rise, fall further, or simply hold once that window closes; nothing in the source indicates which. The documentation's quick comparison table lists a single price each for GPT-5.5 ($5.00) and GPT-5.4 ($2.50) without labeling which pricing column it reflects, so those two figures should not be read as directly comparable line items to GPT-5.6 Sol's stated input, cached input, and output prices without that caveat in mind.