←── back to feed
/topics/gpt-5-6-sol-and-fable-frontier-models

GPT-5.6 Sol and Fable frontier models

41 items3 sourcesupdated 22d agotrend 0

OpenAI released GPT-5.6 with three variants (Sol, Terra, Luna) optimized for different intelligence-cost tradeoffs, alongside Fable, a frontier model for creative tasks. The company launched ChatGPT Work, an agentic platform for enterprise automation, and demonstrated GPT-5.6 Sol Ultra proving a 50-year-old math theorem using 64 subagents.

  • GPT-5.6 Luna outperforms GPT-5.5 on health reasoning while costing 25x less
  • GPT-5.6 Sol Ultra proved novel solution to 50-year-old math problem in under one hour with 64 subagents
  • ChatGPT Work automates workplace tasks across applications, files, and tools to generate documents, presentations, and websites
  • Fable frontier model generates creative content including game mods (5MB Pirate MMO) and satirical PowerPoints
  • GPT-5.6 now default model in Microsoft 365 Copilot
[BSKY]bluesky23
This was one of those impressive AI thresholds for me. I gave GPT-5.6 Sol in Codex control over my computer, and asked it to win the daily challenge for the game Slay the Spire 2 (randomized factors, so can't cheat). Its a complex game.
@emollick · @emollick.bsky.social · ▲142 · 26d
I really want to get artisanal with my model selection:
@emollick · @emollick.bsky.social · ▲17 · 26d
I gave Fable the code: "take this game and do something incredible with it to make it something very different. Be creative"
@emollick · @emollick.bsky.social · ▲54 · 26d
Incredibly annoying when Fable has a forbidden thought in the middle of a long-running project and kills it.
@emollick · @emollick.bsky.social · ▲99 · 27d
Fable: "an 8th grader puts together a powerpoint about the Great Gatsby but obviously did not read the Great Gatsby. Show me that powerpoint! ;)"
@emollick · @emollick.bsky.social · ▲53 · 27d
This time OpenAI announced a novel math proof for a 50 year old problem using a public model (most of the other big math breakthroughs have been with experimental LLMs). GPT-5.6 Sol Ultra, using 64 subagents in just under one hour.
@emollick · @emollick.bsky.social · ▲105 · 27d
When GPT-5 came out, I created a procedural brutalist city builder as a demo (you can see it at the video's start)
@emollick · @emollick.bsky.social · ▲8 · 27d
Here is an example of a real work use case with Codex and 5.6 Sol
@emollick · @emollick.bsky.social · ▲38 · 28d
Notes on GPT-5.6, which includes some interesting new additions to the API (programmatic tool calling and multi-agent in particular) - plus 18 pelicans for the 6 reasoning levels and 3 new models: simonwillison.net/2026/Jul/9/g...
@simonw · @simonwillison.net · ▲88 · 28d
This is lovely. I had a fun conversation about vastness with a virtual William Sieghart which led to a beautiful reading of Keat's Endymion, and one about landscapes of memory and led to a reading of Hopkin's Windhover.
@emollick · @emollick.bsky.social · ▲25 · 28d
Managers, and their belief in the value of AI & how to use it, determines the value they actually get from AI.
@emollick · @emollick.bsky.social · ▲73 · 28d
Pretty long gap since the last dot on the green line. Google doesn't have a frontier model any more, which feels like a missed opportunity.
@emollick · @emollick.bsky.social · ▲80 · 28d
My big takeaway as an early tester of both is that Sol & Fable represent jumps over previous models & have opened a large gap with the next-best AIs. People will have preferences for one or another, but if you doing any work where better “…
@emollick · @emollick.bsky.social · ▲70 · 29d
I was an early tester of GPT-5.6 Sol. I was asked to not share demos until after launch but it is a very good model
@emollick · @emollick.bsky.social · ▲120 · 29d
I had Fable build another thing I always wanted, a full procedural fantasy kingdom generator with economics, trade routes, population growth, wars, lineages, and occasional dragons. First, I worked with it on a plan, then it made it happen.
@emollick · @emollick.bsky.social · ▲141 · 29d
The overwrought language of Fable is an ongoing problem in software and design projects where the AI is otherwise excellent. It just creeps into little bits of text and toasts and menus, even if you try to instruct otherwise. And once it g…
@emollick · @emollick.bsky.social · ▲53 · 30d
This is a key reason I don’t expect the flow of frontier open weights models to continue indefinitely, or even for very much longer. The gap between open and closed capabilities may soom start to grow, not shrink. www.reuters.com/world/bei…
@emollick · @emollick.bsky.social · ▲106 · 30d
Unsurprising but still big: MTurk is on its way out, killed by AI.
@emollick · @emollick.bsky.social · ▲113 · 30d
Fable: "you need to build a game that is coherent, fun and story driven. There are precisely two controls. A slider, one side is labelled Maximum Potato and the other is Formalware, there are four positions. The other is a dial that goes f…
@emollick · @emollick.bsky.social · ▲43 · 31d
Some really interesting research from Anthropic that AI models have spontaneously developed a workspace that "appears to support the functions associated with conscious access"
@emollick · @emollick.bsky.social · ▲98 · 31d
I can basically guarantee you are not being ambitious enough with the work you are assigning Fable
@emollick · @emollick.bsky.social · ▲126 · 31d
I wonder if Fable might do better at tasks it "likes" than those that it doesn't. If you ask it to do something "Fable-y" it reacts with enthusiasm and promises to overdeliver. Otherwise, it just gets to work. Testing this would be pretty …
@emollick · @emollick.bsky.social · ▲58 · 31d
Deleting this because I found I had left a setting on Claude Code allowing writing to memory files on one of my computers, so I am not sure where this all crept in, but it might be my fault!
@emollick · @emollick.bsky.social · ▲93 · 31d