GPT-6.1 Sol review: API pricing, Work/Codex access, and when Astra still wins

Key takeaways

• GPT-6.1 Sol is OpenAI’s cheaper Sol-class model for coding, computer use, and professional work.
• Standard short-context API: $2 / $0.10 cached / $10 output per 1M tokens (Oct 2026 docs).
• Ships in API, ChatGPT Work, and Codex on Plus+. Not in ordinary Chat at launch.
• Positioned as near-Astra quality at about one-fifth Astra’s standard input/output list price.
• Do not confuse Sol with OpenAI Dots (Astra-powered agents) or with Luna (cheap high-volume tier).

Table of contents

1. What GPT-6.1 Sol is
2. API pricing
3. Where you can use it
4. Sol vs Astra
5. Who should switch
6. FAQ

---

What GPT-6.1 Sol is

OpenAI’s naming is crowded: Astra, Sol, Luna, Dots, Work, Codex. Separate the model from the product shell.

GPT-6.1 Sol is a frontier coding and professional model (API id gpt-6.1-sol) that targets near-GPT-6 Astra performance on agentic coding and computer use at a much lower standard token price.

What that means in practice:

• What it is: the default “serious work” Sol upgrade for agents that burn a lot of tokens.
• What limits it: not in consumer Chat at launch; long-context and Fast modes cost more; still proprietary cloud.
• What it is not: an always-on Dot, and not the cheapest Luna tier for spammy volume.

See also our earlier GPT-6 Astra review if you need the top-tier story.

Last verified: October 2026 against OpenAI API model/pricing docs and the GPT-6.1 Sol tool page.

API pricing

Mode / context	Input / 1M	Cached input / 1M	Cache writes / 1M	Output / 1M
Standard, short context	$2.00	$0.10	$2.50	$10.00
Standard, long context	$4.00	$0.20	$5.00	$15.00
Fast (where offered)	2× Standard	2×	2×	2×
Batch / Flex	50% of Standard	50%	50%	50%

Why cached input matters: coding agents reuse big prompts. $0.10 / 1M cached is the lever that makes Sol attractive versus Astra’s higher cached rates.

For comparison, GPT-6 Astra’s short-context Standard list is about $10 input / $50 output per 1M on OpenAI’s pricing table. Sol is the “same job family, thinner bill” option.

Where you can use it

Surface	GPT-6.1 Sol at launch
OpenAI API (gpt-6.1-sol)	Yes
ChatGPT Work	Yes (Plus+)
Codex (app / CLI)	Yes (Plus+)
ChatGPT Chat	No
Free / Go	No
Enterprise / Edu	Often off until admin enables

If your team only lives in ordinary Chat, Sol will not show up no matter how many DevDay posts you paste into Slack.

Sol vs Astra

	GPT-6.1 Sol	GPT-6 Astra
Job family	Coding, computer use, professional work	Top-tier same family
Standard short in/out	~$2 / $10	~$10 / $50
Cached input (short)	~$0.10	~$1.00
When to pick	Most agent loops after you eval quality	When Sol fails your hardest evals

Bottom line: start evals on Sol. Promote to Astra per task family when quality gaps are real, not when a tweet says Astra is “smarter.”

Who should switch

Good fit:

• Teams running coding agents or computer-use loops on the API
• Plus+ users living in Work/Codex who want better Sol-class quality without Astra bills
• Anyone already caching large system prompts

Poor fit:

• Free/Go ChatGPT users waiting for Chat access
• Workloads that already fail quality bars on Sol in your harness
• Buyers who actually wanted Dots (different product)

Related: GPT-6.1 Sol, ChatGPT, GPT-6 Astra review.

FAQ

What is GPT-6.1 Sol?

GPT-6.1 Sol is OpenAI’s September 2026 Sol-class model for agentic coding, computer use, and professional work. It aims for near-GPT-6 Astra quality at much lower standard API token prices.

How much does GPT-6.1 Sol cost on the API?

Short-context Standard rates are about $2 per 1M input, $0.10 per 1M cached input, $2.50 per 1M cache writes, and $10 per 1M output. Long-context rates are higher. Fast mode is 2x Standard; Batch/Flex are 50% of Standard.

Is GPT-6.1 Sol in ChatGPT Chat?

Not at launch. It is available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu (Enterprise/Edu often admin-gated). Free and Go are excluded from the launch rollout.

When should I still use GPT-6 Astra?

Use Astra when your evals show Sol is not good enough on the hardest agentic or computer-use tasks, or when you explicitly need the top tier and can afford roughly 5x standard input/output prices.

Is GPT-6.1 Sol the same as OpenAI Dots?

No. Sol is a model. Dots are always-on agents powered by GPT-6 Astra. You can use Sol for coding agents without buying into Dots.