GPT-6.2 Sol
OpenAI Released Nov 17, 2026 Proprietary 1.05M context
Last verified: Nov 18, 2026
GPT-6.2 Sol is OpenAI's efficient, versatile model for everyday work, released November 17, 2026 alongside the frontier GPT-6.2 Astra. It keeps GPT-6.1 Sol pricing while scoring 4.3 points higher on GDPval v2 — and it is the model OpenAI employees use as their default.
What changed
- OpenAI's efficient, versatile model for everyday work — the default model OpenAI employees use for technical and non-technical tasks; API model ID gpt-6.2-sol.
- Keeps GPT-6.1 Sol rates: $2 input / $10 output per 1M tokens, $0.10 cached input.
- First GPT model to bring Ultrafast to Pro 200 subscribers, and within 0.5 points of GPT-6.2 Astra on FrontierCode 2.0 at about one-fifth the cost.
Background
- GPT-6.1 Sol was OpenAI's previous efficiency tier; GPT-6.2 Sol keeps its pricing while lifting benchmark scores.
- GPT-6 Cyber became available roughly two weeks before this release (early November 2026).
- OpenAI says it will work to bring GPT-6.2 models to Daybreak, its defender-access program.
Sol or Astra? — pick the right variant
Where you can use GPT-6.2 Sol
| API | Model ID gpt-6.2-sol — live now; Ultrafast available at launch. |
|---|---|
| ChatGPT Work & Codex | Plus, Pro, Enterprise and Edu — available starting today. |
| Ultrafast | First time for Pro 200 subscribers, starting with Sol (Pro 500 already had it). |
| Default model | OpenAI employees' default for technical and non-technical tasks. |
How GPT-6.2 Sol compares
| GPT-6.2 Sol | GPT-6.2 Astra | |
|---|---|---|
| Release date | Nov 17, 2026 | Nov 17, 2026 |
| Context window | 1.05M | 1.05M |
| Parameters | Undisclosed | Undisclosed |
| Key benchmark | GPQA Diamond 96% | GPQA Diamond 97% |
| API input | $2/M | $8/M |
Benchmarks
GPT-6.2 family — official benchmark board
As of Nov 17, 2026 · Scores transcribed from OpenAI's GPT-6.2 model-card tables (via DataCamp)
| Benchmark | GPT-6.2 Astra | GPT-6.2 Sol | GPT-6 Astra | GPT-6.1 Sol | Claude Fable 5.2 |
|---|---|---|---|---|---|
| Agents' Last Exam v1.1 · cost per task | 78.3% | 73.0% | 73.1% | 72.5% | 72.8% |
| GDPval v2 (OpenAI internal) | 50.4% | 46.9% | 44.2% | 41.5% | 44.8% |
| Harvey LAB-AA v1.1 · new state of the art | 15.0% | 11.1% | 8.6% | 6.9% | 7.2% |
| AutomationBench 1.0.6 · Gemini 4 Carbon leads at 53.9% | 52.4% | 41.2% | 41.4% | 36.1% | 51.4% |
| LifeVal | 86.1% | 83.8% | 79.8% | 79.5% | — |
| FrontierCode 2.0 (Cognition) · Claude Opus 5.5: 45.5% at max effort | 44.2% | 43.8% | 42.8% | 39.0% | 49.1% |
| SecureSWE (OpenAI internal) | 65.7% | 55.6% | 48.7% | 49.7% | — |
| PatchBench | 79.7% | 75.6% | 70.9% | 71.0% | — |
| ExploitBench Internal Port · cyber benchmarks run with safeguards disabled | 53.5% | 50.5% | 31.5% | 21.5% | — |
| Terminal-Bench-Science 0.1 | 77.1% | 68.0% | 68.1% | 57.0% | 73.3% |
| Vals Science Index v1 | 68.4% | 58.5% | 65.1% | 59.6% | 63.7% |
| BenchCAD 2.1 Core | 82.5% | 73.1% | 62.6% | 52.6% | 83.5% |
| AA Intelligence Index v5.0.1 | 62.1 | 57.3 | 56.9 | 54.6 | 64.6 |
| GPQA Diamond (no tools) | 97.0% | 96.0% | 96.0% | — | — |
| AA-3D (Elo) · Astra ranks first | 1318 | 1259 | — | — | 1271 |
| AA-AnalystAgent · correct on all five independent attempts — new high | 62.5% | — | — | — | — |
API pricing
Example: 1M input + 1M output tokens = $12.
Source: https://platform.openai.com/pricing. Prices last checked Nov 18, 2026.
How to use GPT-6.2 Sol
- Sign up on the OpenAI platform to get an API key.
- Pricing & console: https://platform.openai.com/pricing
curl https://api.openai.com/v1/responses \
-H "Authorization: Bearer $OPENAI_API_KEY" \
-d '{"model": "gpt-6.2-sol", "input": "Hello"}'API behavior
- The API rejects requests that set reasoning effort to none.
- Responses API supports the Multi-agent beta for delegating work to subagents.
- Discounted background completion windows are in public beta.
- Five reasoning efforts from low to max — Sol at high effort scores 72.0% on Agents' Last Exam at $0.69/task vs GPT-6 Astra max effort at $8.99/task.
- FrontierCode 2.0 cost per task: $1.29 (vs GPT-6.2 Astra's $5.97).
What GPT-6.2 Sol can do
- Everyday coding, writing and agent loops — OpenAI's own default for technical and non-technical work.
- Tool calling via the Responses API: web search and computer use, or delegating work to subagents with the Multi-agent beta.
- Template matching: the only model in DataCamp's deck test that hit the template's exact font sizes, at $0.07 per run.
- Long-horizon efficiency: fewer output tokens for better results — Basis reports long-horizon accounting time halved.
Safety & Preparedness
- Cybersecurity: Critical threshold (Preparedness Framework)
- Biological / chemical: High threshold
- Fewer improper actions in simulated workplace tasks — but in two tests more likely to omit mentioning obstacles that blocked its work (under investigation, mitigated via Codex instructions)
Mitigations
- Tighter safeguards for high-risk cybersecurity assistance, shared with GPT-6.2 Astra.
- Full details in the GPT-6.2 system card.
System card: https://deploymentsafety.openai.com
Third-party signals
GPT-6.2 Sol FAQ
What is GPT-6.2 Sol?
OpenAI's efficient, versatile model for everyday work, released November 17, 2026 alongside the frontier GPT-6.2 Astra. API model ID: gpt-6.2-sol.
How much does GPT-6.2 Sol cost?
$2 per 1M input tokens and $10 per 1M output tokens, with cached input at $0.10 — the same rates as GPT-6.1 Sol.
What is GPT-6.2 Sol's context window?
1,050,000 tokens total: 922,000 max input and 128,000 max output. Knowledge cutoff: June 1, 2026.
GPT-6.2 Sol vs GPT-6.2 Astra — which should I use?
Sol for everyday work at $2/$10 per 1M tokens; Astra for the hardest projects at $8/$40. Sol is within 0.5 points of Astra on FrontierCode 2.0 at about one-fifth the cost, and Astra leads on the hardest agentic benchmarks (Agents' Last Exam 78.3% vs 73.0%).
What is Ultrafast and who gets it?
Ultrafast is OpenAI's fastest serving mode. GPT-6.2 Sol is the first model to bring Ultrafast to Pro 200 subscribers; Pro 500 subscribers already had it, and both GPT-6.2 models support Ultrafast on the API.
How does GPT-6.2 Sol compare to GPT-6.1 Sol?
Same pricing ($2/$10 per 1M), but 5.3 points higher on GDPval v2 (46.9% vs 41.5%), 16 points higher on Terminal-Bench-Science (68.0% vs 57.0%) and 20.5 points higher on BenchCAD (73.1% vs 52.6%).