GPT-6 Astra replaced GPT-5.6 Sol as OpenAI's top flagship in September 2026. GPT-5.6 Sol remains on the API at a promotional discount, which makes it the cheaper way to get a previous-generation OpenAI flagship.
GPT-5.6 Sol is 2.5× cheaper than GPT-6 Astra on input and output ($4.00/$20.00 vs $10.00/$50.00 per 1M). That price includes a 20% promotional cut that OpenAI lists through at least Nov 21, 2026, so budget for a higher rate afterwards.
| GPT-6 Astra | GPT-5.6 Sol | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Tier | Flagship | Flagship |
| Input / 1M tokens | $10.00 | $4.00 |
| Output / 1M tokens | $50.00 | $20.00 |
| Cache hit / 1M | $1.00 | $0.40 |
| Cache write / 1M | $12.50 | $5.00 |
| Context window | 1M | 1M |
| Notes | New top flagship (Sep 3, 2026); rolling out to Plus/Pro/Business/Enterprise; >272K tokens: $20/$75, cache $2 | Cut 20% Aug 21 (promo thru Nov 21); >272K tokens: $8/$30, cache $0.80 |
GPT-5.6 Sol is 2.5× cheaper on input tokens ($4.00 vs $10.00 per 1M). GPT-5.6 Sol is 2.5× cheaper on output tokens ($20.00 vs $50.00 per 1M).
| Workload | GPT-6 Astra | GPT-5.6 Sol | Difference |
|---|---|---|---|
| Support chatbot 10,000 conversations × (2,000 input + 400 output tokens) | $400 | $160 | GPT-5.6 Sol saves $240 |
| Document Q&A 1,000 questions × (30,000 input + 800 output tokens) | $340 | $136 | GPT-5.6 Sol saves $204 |
| Coding agent session 4M input tokens (90% cache hits) + 150,000 output tokens | $15.10 | $6.04 | GPT-5.6 Sol saves $9.06 |
Standard rates per 1M tokens from data on this site. The coding-agent row assumes 90% of input is read from the prompt cache (full input price where a model has no cache rate). Cache-write fees, batch discounts and long-context surcharges listed in the notes are not included. Updated November 2, 2026.
you want OpenAI's newest flagship and your evals show it solving tasks GPT-5.6 Sol cannot, enough to justify a 2.5× token bill.
you have prompts validated on the 5.6 generation, want flagship-class output at a lower price, and can re-check costs once the promotion ends.
GPT-5.6 Sol's rate reflects a 20% cut made on Aug 21, 2026 as a promotion; OpenAI's pricing page says it runs at least through Nov 21, 2026. If the promotion lapses, 5.6 Sol's advantage over Astra shrinks from 2.5× to 2×. Anyone committing budgets past late November should model both.
Both step up above 272K input tokens: Astra to $20/$75 with $2 cache hits, 5.6 Sol to $8/$30 with $0.80. Cache hits at base tier are $1.00 on Astra versus $0.40 on 5.6 Sol, so the 2.5× ratio holds for cached input too.
Within the GPT-6 family, GPT-6.1 Sol is listed at $2.00/$10.00 per 1M with $0.10 cache hits — cheaper than GPT-5.6 Sol even at its promo price. If you are moving off 5.6 Sol for cost reasons, compare it before Astra. OpenAI's Batch API halves all three.
Yes, 2.5× cheaper on input, output and cache hits at current promotional pricing.
OpenAI lists the promotional pricing as available at least through November 21, 2026.
Yes: Astra $20/$75 and GPT-5.6 Sol $8/$30 per 1M for prompts over 272K tokens.
How we choose: prices come from each vendor's official pricing page and are re-checked every week. We take no affiliate fees, so rankings aren't for sale.