GPT-5.6 Sol
OpenAI · Released 9 Jul 2026 · gpt-5.6-sol
$4.00
Input price / 1M tokens
#50 of 61, cheapest first
$20.00
Output price / 1M tokens
#50 of 61, cheapest first
1.05M
Context window
#2 of 71
79
Output tokens / second
#41 of 61
64.6%
SWE-bench Pro
#2 of 16
72.7%
DeepSWE v1.1
#6 of 17
88.8%
Terminal-Bench 2.1
#3 of 17
62.6%
OSWorld 2.0
#5 of 10
Summary
GPT-5.6 Sol is a proprietary model from OpenAI, released on 9 Jul 2026.
At $4.00 input and $20.00 output per million tokens, it is #50 of 61 on input price, cheapest first.
Its 1.05M-token context window ranks #2 of 71.
Measured output speed is 79 tokens per second, #41 of 61.
Flagship of the GPT-5.6 family, for complex professional work, coding and long-horizon agentic tasks. It adds max reasoning effort and Pro mode.
Benchmarks · 8 reported
| Benchmark | Score | Bar | Rank | Source |
|---|---|---|---|---|
| SWE-bench Pro | 64.6% | #2 of 16 | OpenAI launch postSelf-reported | |
| DeepSWE v1.1 | 72.7% | #6 of 17 | OpenAI launch postSelf-reported | |
| Terminal-Bench 2.1single agent (91.9% with ultra multi-agent) | 88.8% | #3 of 17 | OpenAI launch postSelf-reported | |
| OSWorld 2.0 | 62.6% | #5 of 10 | OpenAI launch postSelf-reported | |
| BrowseCompsingle agent (92.2% with ultra multi-agent) | 90.4% | #2 of 7 | OpenAI launch postSelf-reported | |
| GPQA Diamond | 94.6% | #2 of 27 | OpenAI launch postSelf-reported | |
| FrontierMath Tier 4 (v2) | 83% | – | OpenAI launch postSelf-reported | |
| Agents' Last Examresults table value | 52.7% | – | OpenAI launch postSelf-reported |
Self-reported means the lab ran the test itself. A rank appears only where several labs report the same version of a benchmark.
Similar models
Models with the most benchmarks in common with GPT-5.6 Sol, and the closest scores.
SWE-bench Pro
#2 of 16Harder multi-file engineering tasks
- GPT-5.6 Sol64.6
- GPT-5.6 Terra63.4
- GPT-5.6 Luna62.7
DeepSWE v1.1
#6 of 17Long coding tasks in real repos
- GPT-5.6 Sol72.7
- GPT-5.6 Terra69.6
- GPT-5.6 Luna67.2
- GPT-6 Astra74.1
- Gemini 3.8 Flash73.7
- Kimi K367.5
Terminal-Bench 2.1
#3 of 17Terminal tasks, earlier set
- GPT-5.6 Sol88.8
- GPT-5.6 Terra87.4
- GPT-5.6 Luna84.7
- Gemini 3.8 Flash89.4
- Kimi K388.3
OSWorld 2.0
#5 of 10Long desktop and web workflows
- GPT-5.6 Sol62.6
- GPT-5.6 Terra50.2
- GPT-5.6 Luna45.6
- GPT-6 Astra72.6
- Gemini 3.8 Flash59
Compare GPT-5.6 Sol side by side
Pre-filled with the two closest models. Swap any of them.
Sources
- current price, promo end date, tiers, long-context rule: developers.openai.com/api/docs/pricing, checked 2 Oct 2026
- context (max input 922K), max output, knowledge cutoff (Feb 16 2026), modalities, api id, promo text: developers.openai.com/api/docs/models/gpt-5.6-sol, checked 2 Oct 2026
- release date (Jul 9 2026), price cut on Aug 21 2026: developers.openai.com/api/docs/changelog, checked 2 Oct 2026
- launch price $5/$30, cache write 1.25x, 30-minute cache life, benchmarks: openai.com/index/gpt-5-6/, checked 2 Oct 2026
- output speed: artificialanalysis.ai/models/gpt-5-6-sol, checked 2 Oct 2026
Building on GPT-5.6 Sol?
Get the architecture right before the bill arrives.
We size caching, routing and fallbacks for your workload on a free call, and tell you where a cheaper model is good enough.