DEV Community

IA Compa
IA Compa

Posted on Originally published at iacompa.com

GPT-6.2 Sol: OpenAI's everyday-work model scores 57.3 on the Intelligence Index at $2/$10 per million tokens

OpenAI released GPT-6.2 Sol on November 17, 2026, for everyday coding, writing and agent workflows. It is available in ChatGPT Work, Codex and the API as gpt-6.2-sol.

As of November 17, 2026, GPT-6.2 Sol ranks #5 among models on the Artificial Analysis Intelligence Index (57.3 at max effort) and keeps GPT-6.1 Sol's $2/$10 API prices: flagship-class scores at a quarter of GPT-6.2 Astra's token price.

What changed

Standard API rates per million tokens are $2 input, $0.10 cached input and $10 output. Like Astra, Sol accepts text and images and returns text. OpenAI documents tool calling through the Responses API.

On November 17, 2026, Artificial Analysis recorded a 57.3 Intelligence Index score at max effort and ranked Sol #5 in that snapshot. OpenAI's release also reported that Sol was within 0.5 points of GPT-6.2 Astra on Cognition's FrontierCode 2.0 at about one-fifth the cost in that evaluation, with both at max effort. The benchmark checks whether code changes meet maintainers' standards for merging.

Sol's API provides a 1,050,000-token context window, with up to 922,000 input and 128,000 output tokens. OpenAI lists higher pricing above 272,000 input tokens. Sol also brings Ultrafast to ChatGPT Pro 200 at launch. Pro 500 can use Ultrafast with both models. The API offers Ultrafast for both at separate rates.

Metrics

Metric GPT-6.2 Sol
Artificial Analysis Intelligence Index (max) 57.3 (v5.0.1)
Output speed (max) 85 tokens/s
Input price per 1M tokens $2.00
Output price per 1M tokens $10.00
Context window 1.05M tokens
Cost per Artificial Analysis Intelligence Index task (max) $1.01

Reasoning effort

Artificial Analysis tests this model at several reasoning-effort settings. Higher effort usually scores higher but costs more and answers more slowly.

Effort Intelligence Index Output speed First token
Max 57 85 tokens/s 58 s
Extra high 56 80 tokens/s 19 s
High 55 78 tokens/s 5.6 s
Medium 53 74 tokens/s 2.7 s
Low 41 71 tokens/s 1.7 s

Price: $2 / $10 per 1M input / output tokens at every effort.

Efforts available: low, medium (default), high, xhigh, max, and ultra (only in Codex CLI / Codex app / ChatGPT Work). It supports neither none nor minimal: the API rejects requests that set reasoning effort to none, and OpenAI's GPT-6.2 guide tells anyone migrating from none or minimal to use low instead. Tool calling requires the Responses API. New ChatGPT Work chats start with GPT-6.2 Sol at Light on supported consumer plans, or Medium in Enterprise workspaces. In Codex and ChatGPT Work its context window is up to 1 million tokens. Ultra is not an API value: it runs at maximum reasoning and lets the model delegate parts of the task to parallel subagents. Ultrafast is a speed tier (service_tier), not an effort level.

Benchmarks & resources

Why it matters

Test Sol on routine code reviews, customer support drafts or regularly run agent jobs. Compare completion rates and cost on your own examples. Use the results to identify which cases justify the higher standard token rates of GPT-6.2 Astra.

Updates

  • November 17, 2026 — Artificial Analysis has published measured performance for GPT-6.2 Sol: at max effort it averages 7.9 minutes per Intelligence Index task and 57.55 seconds to the first answer token. Sol scores 3 points below GPT-6.1 Sol on the Coding Agent Index, and its AA-Omniscience hallucination rate at max effort is 55%.

Sources

Originally published at iacompa.com

Top comments (0)