GPT-6.2 Sol
OpenAI's everyday model for coding and agents, at GPT-6.1 Sol's $2/$10 price
Context window
1.05M
Input / 1M tokens
$2.00
Output / 1M tokens
$10.00
Provider
OpenAI
Same price as GPT-6.1 Sol. Cached input $0.10 per 1M; cache writes $2.50 (30-minute) or $5 (12-hour). Prompts over 272K input tokens cost $4/$15 for the full request (cached $0.20). Batch is half price ($1/$5); the 2-hour and 8-hour background completion windows (Responses API, public beta) are 30% and 40% below standard. Fast mode $4/$20. Ultrafast is 3x standard ($6/$30) through February 28, 2027, then 6x ($12/$60). No Flex tier. Regional (US or EU) processing adds 10%. Max input 922K tokens, max output 128K; knowledge cutoff June 1, 2026. · Data verified 2026-11-20
GPT-6.2 Sol (November 17, 2026) is OpenAI's everyday model for coding, writing and agents at scale, and the successor to GPT-6.1 Sol at the same $2/$10 per 1M tokens. It has a 1.05M-token context window (922K max input, 128K max output), accepts text and images, and has a knowledge cutoff of June 1, 2026. OpenAI says it finishes tasks with fewer output tokens, so paid ChatGPT plans complete more tasks than with GPT-6.1 Sol, and that at max effort it came within half a point of GPT-6.2 Astra on FrontierCode 2.0 (43.8% vs 44.2%) at about one-fifth the cost. VentureBeat counts it ahead of GPT-6.1 Sol on eight of nine OpenAI launch charts, but Artificial Analysis measured it 3 points lower than GPT-6.1 Sol on its Coding Agent Index, driven by a 14-point drop on SWE-Atlas-QnA-AA. It is the default for new ChatGPT Work chats on consumer plans and the first model to get Ultrafast on the $200 Pro plan.
How to access it
OpenAI API since November 17, 2026 (model ID gpt-6.2-sol) on the Responses, Chat Completions and Batch endpoints, with Fast and Ultrafast processing; tool calling requires the Responses API. In ChatGPT it is in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu users (Enterprise and Edu admins must enable it). New Work chats default to Sol at Light effort on consumer plans and at Medium in Enterprise workspaces. Ultrafast with Sol is available on the $200 and $500 Pro plans and in Enterprise and Edu workspaces where an owner turns it on. Also on Vercel AI Gateway (openai/gpt-6.2-sol).
Strengths
- ✓73.0% on Agents' Last Exam v1.1 vs 72.5% for GPT-6.1 Sol and 72.8% for Claude Fable 5.2, at $1.32 per task (OpenAI-reported)
- ✓43.8% on FrontierCode 2.0 at max effort, within half a point of GPT-6.2 Astra at about one-fifth the cost (OpenAI-reported)
- ✓68.0% on Terminal-Bench-Science 0.1 vs 57.0% for GPT-6.1 Sol (OpenAI-reported)
- ✓Same $2/$10 price as GPT-6.1 Sol, with fewer output tokens per task, per OpenAI
- ✓1.05M-token context window and 128K max output
Best for developers who...
When to choose it (and when not to)
Reach for GPT-6.2 Sol when...
- →Everyday coding, writing and agent work where Astra would cost four times as much
- →You use GPT-6.1 Sol today: same price, and higher on eight of nine OpenAI launch charts
- →Long-running or high-volume workloads where cost per task matters
Look elsewhere if...
- ✕The hardest projects and science work, where GPT-6.2 Astra leads by several points on most of OpenAI's charts
- ✕Coding agents tuned on GPT-6.1 Sol before you test them: Artificial Analysis measured a 3-point drop on its Coding Agent Index
- ✕Very cheap, high-volume tasks (GPT-6.1 Terra costs $1/$5 and GPT-6 Luna $0.10/$0.50)
- ✕Code that sets temperature, top_p, logprobs or reasoning effort none, which these models reject
How to use it
- ›Start at medium reasoning effort; if older code used none or minimal, move to low (OpenAI GPT-6.2 guide)
- ›Use the Responses API for tool calling
- ›Raise effort to max for hard coding: that is the setting where OpenAI reports Sol near Astra on FrontierCode 2.0
Quickstart
Pythonfrom openai import OpenAI
client = OpenAI()
response = client.responses.create(
model="gpt-6.2-sol",
reasoning={"effort": "medium"},
input="Review this pull request diff and list the risky changes: ..."
)
print(response.output_text)Install with pip install openai. OpenAI's GPT-6.2 guide suggests starting at medium effort; tool calling needs the Responses API.
API model id: gpt-6.2-sol
Benchmarks
| Benchmark | Score | Notes |
|---|---|---|
| Agents' Last Exam v1.1 | 73.0% | OpenAI launch chart as reported by DataCamp and VentureBeat; GPT-6.1 Sol 72.5%, GPT-6.2 Astra 78.3%, Claude Fable 5.2 72.8% (max); 72.0% at high effort |
| FrontierCode 2.0 | 43.8% (max) | OpenAI launch chart as reported by DataCamp and VentureBeat; GPT-6.1 Sol 39.0%, GPT-6.2 Astra 44.2% at about five times the cost per task ($5.97 vs $1.29) |
| GDPval v2 | 46.9% | OpenAI internal evaluation, as reported by VentureBeat; GPT-6.1 Sol 41.5% |
| Harvey LAB-AA v1.1 | 11.1% | OpenAI launch chart as reported by DataCamp and VentureBeat; GPT-6.1 Sol 6.9%, Claude Fable 5.2 7.2% |
| Terminal-Bench-Science 0.1 | 68.0% | OpenAI launch chart as reported by DataCamp and VentureBeat; GPT-6.1 Sol 57.0% |
| AutomationBench 1.0.6 | 41.2% | OpenAI launch chart as reported by DataCamp and VentureBeat; GPT-6.1 Sol 36.1% |
| Vals Science Index v1 | 58.5% | OpenAI launch chart as reported by DataCamp and VentureBeat; GPT-6.1 Sol 59.6%, the one chart where Sol trails its predecessor |
| Artificial Analysis Intelligence Index v5.0.1 | 57.3 | Per OpenAI's launch chart; GPT-6.1 Sol 54.6. Artificial Analysis reports 57, up 2 points, level with GPT-6 Astra |
| GPQA Diamond | 96.0% (no tools) | OpenAI GPT-6.2 system card; GPT-6.2 Astra 97.0% |
| SEC-bench Pro | 88.34% | OpenAI GPT-6.2 system card, pass@1; GPT-6.1 Sol 78.8% |
Tools powered by GPT-6.2 Sol
Compare GPT-6.2 Sol
GPT-6.2 Sol vs Claude Opus 5.5
Anthropic - 1M ctx
GPT-6.2 Sol vs Gemini 4 Neon
Google - 1M (shared input + output) ctx
Compare GPT-6.2 Sol with any other model
Build a comparison →All model comparisons →