Tasks and inputs
GPT-6.2 Sol is listed for Image-text-to-text, Text generation, Coding & Code Generation, Computer Use. It accepts text, image and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
GPT-6.2 Sol is OpenAI's everyday model for coding, agents and writing: $2/$10 per million tokens, 1.05M context, released 17 November 2026.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Pricing conditions: Indicative Standard pricing: $2 input / $0.10 cached / $10 output per 1M; cache writes $2.50 (30-min) or $5 (12-h). Over 272K input: $4 / $0.20 / $15. Batch 50% off; fast 2x; Ultrafast 3x until 28 Feb 2027 then 6x; background 30-40% off (beta). Confirm on your OpenAI billing page.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
Same price, new model. GPT-6.2 Sol keeps the $2 input and $10 output per million tokens that GPT-6.1 Sol charged, so for most teams the upgrade is a one-line change of model ID rather than a budget meeting. OpenAI calls it the go-to model for everyday work, from code to agents running at scale.
gpt-6.2-solPrices are indicative. Fast mode doubles them, and Ultrafast triples them until the end of February 2027, then rises to six times. Check your own account before committing.
The interesting comparison is upward, not backward. On FrontierCode 2.0, OpenAI reports that Sol finishes within half a point of its expensive sibling, GPT-6.2 Astra, at about a fifth of the cost per task. That is a publisher-reported result and worth testing rather than trusting. Independent coding indexes published after launch tell a more modest story: Sol's coding score dipped slightly against 6.1 Sol, pulled down by one question-answering benchmark, while its overall intelligence score rose two points.
So which do you believe? Your own repository.
GitHub, which put Sol into Copilot on day one, reports early testing where it flagged nearly twice as many real issues in code review as GPT-6.1 Sol. Code review may be where the gain is most visible.
Tool calling requires the Responses API, where Sol gets the full hosted set: web and file search, code interpreter, hosted shell, apply_patch, computer use, MCP, tool search, skills and image generation. Reasoning cannot be turned off; an effort of none is rejected. Fine-tuning is not offered. Rate limits match Astra, starting at 5,000 requests per minute.
In ChatGPT the model sits in ChatGPT Work and Codex for paid tiers, and Pro 200 subscribers get Ultrafast starting with Sol. In GitHub Copilot it is available on Pro+, Max, Business and Enterprise.
Replay a day of logged traffic from 6.1 Sol against 6.2 Sol. Compare three things: review findings that a human confirms, tokens per finished task, and any regressions on retrieval-style questions. If those hold, flip the default. If you also run GPT-6 Luna for cheap traffic, check whether Sol's efficiency closes enough of the gap to simplify routing. The full OpenAI catalogue lists the alternatives.
Compare the model on the work you need to complete.
GPT-6.2 Sol is listed for Image-text-to-text, Text generation, Coding & Code Generation, Computer Use. It accepts text, image and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
Documented limits: 1.05M context and 128K maximum output. The listed route has starting rates of $2.00 input and $10.00 output per million tokens. Measure a completed task using the prompts, documents and tools your application will use. Include reasoning and retries in the result.
Use the model identifier and account requirements for the selected API route. Confirm its request limits and output format with one complete task before connecting a larger workflow.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
OpenAI
GPT-6.2 Astra is OpenAI's November 2026 flagship for hard professional, scientific and agentic work, at $8/$40 per million tokens with a 1.05M context.
Tencent
Tencent Hy4 Flash is the cost-focused companion to Hy4, keeping vision and agent skills at a far lower token price. Availability is still unsettled.
Alibaba
Qwen4 Max (qwen4-max) is Alibaba's first Qwen4-generation flagship: 1M-token context, text, image and video input, thinking on by default, $1.50/$4.50 per 1M.
Sarvam
Sarvam-105B-1711 is a refreshed open post-train of Sarvam's 105B MoE, tuned for tool calling, instruction following, agents and Indian languages.
IBM
Granite 4.1 8B LLMON Preview is IBM's model trained on block-structured LLMON prompts to resist indirect prompt injection in agents and RAG.
NVIDIA
Nemotron-Labs Sigma 8B is NVIDIA's open continuous diffusion language model, denoising 64-token blocks in parallel; English, GPU-only, 8K context.