Skip to main content

September 30, 2026 · Brad Gessler

Claude Opus 5.5, GPT-6.1 Sol and seven more models

GBrain workspaces that answered with Opus 4.8 or Opus 5 now answer with Opus 5.5, which costs a fifth less than both. GPT-6.1 Sol costs less than GPT-5.5.

GBrain now offers nine new models from Anthropic and OpenAI, and Claude Opus 5.5 is the one a new workspace starts on. A model is the AI that writes the answers; each GBrain workspace uses one, and an admin can pick another.

Four of them cost less than the models they replace

Claude Opus 5.5 is the model new workspaces start on, and every workspace that ran Opus 4.8 or Opus 5 now runs it. Anthropic says it performs at the level of Claude Fable 5.1 on most work. Models are priced by the token, a piece of a word: Opus 5.5 costs $4 per million tokens it reads and $20 per million it writes, a fifth less than Opus 4.8 and Opus 5 ($5 and $25).

OpenAI describes GPT-6.1 Sol as built for complex coding and professional work, at a lower cost than GPT-6 Astra. At $2 to read and $10 to write, it costs less than GPT-5.5 ($5 and $30) on both, and it is now listed first among OpenAI’s models.

GPT-6 Luna is the lowest-priced GPT-6 model, at $0.10 to read and $0.50 to write, for large amounts of work.

Claude Sonnet 5.5 runs 30% faster than Sonnet 5 and costs up to 30% less, by Anthropic’s measure.

Two models for the hardest work

Claude Fable 5.1 and GPT-6 Astra are the most capable models GBrain offers from each company. They cost more than the models above.

Every new model

ModelFromPer million tokens, read / written
Claude Opus 5.5Anthropic$4 / $20
Claude Fable 5.1Anthropic$10 / $50
Claude Sonnet 5.5Anthropic$2 / $10
GPT-6.1 SolOpenAI$2 / $10
GPT-6 AstraOpenAI$10 / $50
GPT-6 SolOpenAI$2 / $10
GPT-6 LunaOpenAI$0.10 / $0.50
GPT-5.6 TerraOpenAI$2 / $12
GPT-5.6 LunaOpenAI$0.20 / $1.20

Each model’s page also has the lower price for text the model has read recently and can reuse. “GPT-5.6” is now listed by its full name, GPT-5.6 Sol.

What you might notice

Answers from Opus 5.5. A workspace that ran Opus 4.8 or Opus 5 now answers with Opus 5.5. A workspace that chose any other model keeps it.

What you need to do

Nothing. An admin, the person who runs the workspace’s settings, can choose a different model in the workspace’s Settings, under Model.

The announcements are Anthropic’s posts on Claude Opus 5.5, Claude Sonnet 5.5 and Claude Fable 5.1, and OpenAI’s model changelog.

Set up a GBrain workspace

Spin one up in a few minutes, and every AI the team uses shares one memory and the same connected accounts.