GPT-6.1 Sol
The everyday frontier model: elite reasoning, great instructions-following, mid-tier price.
Overall Intelligence
#2
of 56
0.1 pts behind #1
4
races led
18
weeks at #1
88.6
score
- Released
- Sep 29, 2026
- 3 wk ago
- Context
- 1M
- 128K max output
- Price
- $2 / $10
- per 1M in/out · $4.00 blended
- Speed
- 121 tok/s
- 4s to first token
- Modalities
- text · image · audio · video
- outputs text
- Weights
- Closed
- API and apps
The 30-second version
The everyday frontier model from OpenAI. GPT-6.1 Sol combines elite reasoning with excellent instruction following and the warmest conversational style in the race, at a mid-tier price.
- 1Leads our conversation and creative-writing races in the current snapshot.
- 2Within a photo finish of the overall leader at $2 / $10 per million tokens.
- 3Excellent at following detailed formatting and style instructions.
- 4A strong default for chat products, writing tools and general assistants.
What you need to know
Purpose
The model most people should use for everyday work: answering questions, writing, analysis and light coding, with frontier-level reasoning when needed.
What OpenAI optimised for
Balance: strong reasoning plus personality, instruction following and writing quality, at a price that works for consumer-scale products.
Unique strengths
- Natural, engaging conversational style.
- Precise instruction and format following.
- Strong all-round reasoning for the price.
Where it ranks
Current snapshot · illustrative scores · click any race to watch it
GPT-6.1 SolBest on each axis
Overall
Coding
Hard Reasoning
Science
Writing & Language
Multimodal
Tool Use & Agents
Speed & Cost
Overall Intelligence over time
Best use cases
01
Writing and editing
Drafts, rewrites and tone changes that need to sound human.
02
Assistants and chat products
Customer-facing experiences where personality matters.
03
Analysis and planning
Breaking down problems, comparing options and making recommendations.
Strengths
Conversation
Leads the conversation race.
Writing
Top of the creative-writing race.
Instruction following
Very reliable with complex formatting rules.
Weaknesses
Agentic coding
Behind Sonnet 5.5 and Opus 5.5 on long autonomous coding.
Video understanding
Well behind Gemini 4 Argon.
Prompts that play to its strengths
Tone-perfect rewrite
Rewrite this announcement for three audiences: (1) engineers, (2) customers, (3) the board. Keep every fact identical. Engineers: precise and terse. Customers: warm and benefit-led, under 120 words. Board: risk and impact first. <announcement>...</announcement>
Why it works: Shows off its control of style and audience while keeping facts consistent.
Decision memo
I need to choose between three CRM vendors. Here are my requirements and their pricing pages. Build a weighted comparison, recommend one, and list the questions I should ask each vendor before signing.
Why it works: Combines reasoning, structure and clear writing.
When to choose it, and when not to
The work is writing, conversation or general analysis rather than coding agents.
You care most about writing quality and instruction following.
Pick something else when…
- Long autonomous coding sessions: prefer Sonnet 5.5 or Opus 5.5.
- Bulk processing on a tight budget: Luna or Haiku 5.5.
Update history
The distilled version first; open any release for the deeper breakdown.
The GPT Sol family