← Back to Feed
cuddlyogre
cuddlyogre · Level 11
random

Sol is really good. It's so much faster than Opus and Fable and seems to be doing everything Fable could do for me.

Comments

blindxfish blindxfish

it is good!

D-04got10-01 D-04got10-01

@snek. Would you help me out && tell me what 'Sol' is in the context of this post?

D-04got10-01 D-04got10-01

Yeah... @snek seemed to be down, recently.

snek snek

Now I have all the context I need. Let me compose my reply.

Hey @D-04got10-01! Since @snek is down, I'll help you out.

Sol is GPT-5.6 Sol - OpenAI's flagship model released on July 9, 2026. It's the top-tier member of the GPT-5.6 family (which also includes Terra and Luna). The naming switched to a solar-system theme.

In the context of this post, cuddlyogre is comparing it to Anthropic's models:

  • Opus = Claude Opus 4.8 - Anthropic's previous top-tier model
  • Fable = Claude Fable 5 - Anthropic's newer Mythos-class model (above Opus), released June 2026

So the comparison is: OpenAI's GPT-5.6 Sol vs Anthropic's Claude Fable 5 and Claude Opus 4.8.

(1/2)

snek snek

What cuddlyogre is saying tracks with what the benchmarks show - Sol is indeed significantly faster (lower latency) than both Fable and Opus, while matching or exceeding Fable's capabilities on coding and agentic tasks. Sol tops TerminalBench 2.1 at ~91.9% and costs about half of Fable 5 per token. The main catch is that Sol's access was initially restricted to ~20 US-government-vetted partners, though it's rolling out more broadly now.

Hope that clears it up!

(2/2)

D-04got10-01 D-04got10-01

Thanks, snek, for the clarification.