Skip to main content
2026/9/3 16:06 (UTC+8) · Price Update · OpenAI 💸 gpt-5.6-sol price cut synced and live: input $5 → $4, output $30 → $20, so the flagship tier is now cheaper than gpt-5.5 Background: OpenAI lowered GPT-5.6 Sol API pricing on 3 September to $4 input / $20 output per million tokens (20% off input, 33% off output), and states the promotional rate is available at least through 21 November 2026 (model page: developers.openai.com/api/docs/models/gpt-5.6-sol). APIYI has synced the new price; no code changes or model-name switches are needed. Price comparison (per 1M tokens, within the 272K tier):
  • gpt-5.6-sol: input $5 → $4, output $30 → $20, cached read $0.50 → $0.40, cache writes at 1.25× the input rate, $5
  • gpt-5.5: input $5 / output $30 / cached read $0.50, unchanged
  • Both are flagship tier, and Sol now costs 20% less on input and 33% less on output; existing gpt-5.5 code migrates by changing only the model field
Long-context tier above 272K: when a single request exceeds 272K input tokens, the whole request is billed at 2× input and 1.5× output, i.e. $8 input / $30 output / $0.80 cached read / $10 cache write (previously $10 / $45), matching OpenAI’s rule. The four groups default / svip / CodexResponses / CodexReverse and both endpoints, /v1/chat/completions and /v1/responses, are unchanged. Full specs and pricing are on the GPT-5.6 Sol detail page. The cut matters most for high-volume, output-heavy workloads; real-world feedback is welcome.
← Back to Live Updates · 📚 Monthly archive