Closed SourceOpenAIReleased 2026-09

GPT-6 Sol

Cost-efficient mid-tier model for complex professional work and coding

Context window

1.05M

Input / 1M tokens

$2.00

Output / 1M tokens

$10.00

Provider

OpenAI

Data verified 2026-09-24

GPT-6 Sol is OpenAI's balanced model in the GPT-6 series, trained with the same methods as GPT-6 Astra. It brings Astra-level capabilities in factuality, coding, and professional tasks at significantly lower cost through improved inference and caching. Designed for demanding work including agentic coding, business automation, and software engineering.

Capability index

Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.

Coding
8
Reasoning
8
Math
7
Multimodal
0
Long context
8
Speed
7
Cost efficiency
9

How to access it

Available via OpenAI API. Also available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users.

Strengths

  • 50% price reduction vs. GPT-5.6 Sol while maintaining comparable capability
  • Halved factual error rate compared to predecessor
  • Strong on coding and computer use tasks
  • Astra-level factual reliability at much lower cost

Best for developers who...

Complex coding and agentic workflowsBusiness workflow automationSoftware engineering and code reviewProfessional knowledge work requiring iteration

When to choose it (and when not to)

Reach for GPT-6 Sol when...

  • Need Astra-level capabilities at mid-tier pricing
  • Running at scale with cost constraints
  • Balancing performance and affordability

Look elsewhere if...

  • Requiring absolute maximum performance (use Astra instead)
  • Ultra-cost-sensitive work (use Luna instead)

How to use it

Quickstart

Python
from openai import OpenAI
client = OpenAI()
response = client.chat.completions.create(
  model="gpt-6-sol",
  messages=[{"role": "user", "content": "Hello"}]
)
print(response.choices[0].message.content)

Use service_tier='fast' for priority processing at 2x the standard rate

API model id: gpt-6-sol

Benchmarks

BenchmarkScoreNotes
AutomationBench33.2%At maximum reasoning level, outperforms Claude Opus 5 at max effort at 9% of cost per task
Internal Factuality Evaluation~50% error reductionBased on de-identified real ChatGPT conversations; half the errors of GPT-5.6 Sol

Source: OpenAI Official Announcement

Compare GPT-6 Sol

Compare GPT-6 Sol with any other model

Build a comparison →
All model comparisons →

Learn the concepts