Claude Opus 4.8 launches, scoring 69.2% on SWE-bench Pro to outperform GPT-5.5 and Gemini 3.1 Pro
It introduces a "Dynamic Workflows" preview for parallel subagents.
Opus 4.8 scores 69.2% on SWE-Bench Pro, 10 points higher than GPT-5.5.
Most interesting part of the release blog is “Dynamic Workflows”:
“This new feature, available in research preview, allows Claude to take on even bigger tasks in Claude Code. Claude can plan the work and then run hundreds of parallel subagents in a single session (and with Opus 4.8, the agents can run for even longer). It then verifies its outputs before reporting back to the user.”
Andrew Curran@AndrewCurran_System card: https://cdn.sanity.io/files/4zrzovbb/website/c886650a2e96fc0925c805a1a7ca77314ccbf4a6.pdf
'Not only that, but we plan to release a new class of model with even higher intelligence than Opus.'
The Mythos release draws near. The rumor for some time is that Claude Mythos will release in about two weeks, mid June.
Andrew Curran@AndrewCurran_Andrew Curran@AndrewCurran_'Not only that, but we plan to release a new class of model with even higher intelligence than Opus.' The Mythos release draws near. The rumor for some time is that Claude Mythos will release in about two weeks, mid June.
I think it's becoming clearer that programmatic sub-agent calling is the way to go over the legacy tool-calling format (which I've been pushing for since RLMs came out)!
I do wonder though if the generated "workflow" looks more eager or compiled (a design decision I've also been unsure about, because it affects how these models are trained to act); dynamic seems to imply the former but the example they give in the blog makes it kind of unclear. either way, scaling the flexibility of subagent deployment without polluting the context of the main Claude Code instance is gonna be huge
Yuchen Jin@Yuchenj_UWOpus 4.8 scores 69.2% on SWE-Bench Pro, 10 points higher than GPT-5.5. Most interesting part of the release blog is “Dynamic Workflows”: “This new feature, available in research preview, allows Claude to take on even bigger tasks in Claude Code. Claude can plan the work and then run hundreds of parallel subagents in a single session (and with Opus 4.8, the agents can run for even longer). It then verifies its outputs before reporting back to the user.”
Claude Opus 4.8 Benchmarks
#980Lisan al Gaib@SCALING01Opus 4.8 is indeed #1 on FrontierSWE
Lisan al Gaib@scaling01Anthropic says Opus 4.8 ranks 1st on FrontierSWE
Lisan al Gaib@scaling01Opus 4.8 is indeed #1 on FrontierSWE
Anthropic says Opus 4.8 ranks 1st on FrontierSWE
not sure if that includes GPT-5.5
and the rankings for the models changed
Lisan al Gaib@scaling01Anthropic says Opus 4.8 ranks 1st on FrontierSWE
#980Lisan al Gaib@SCALING01Claude 4.8 Opus System Card
Lisan al Gaib@scaling01Claude Opus 4.8 Benchmarks
Lisan al Gaib@scaling01Claude 4.8 Opus System Card
@scaling01 Just 5 more versions to reach Mythos, I can't contain my excitement.
Lisan al Gaib@scaling01Claude Opus 4.8 Benchmarks
@AndrewCurran_ good chance that mythos has basically solved all solvable hle tasks or is somewhat close
Andrew Curran@AndrewCurran_HLE.
Big TerminalBench improvements from Opus 4.7 still behind 5.5
Massive knowledge work improvements!
Claude Opus 4.8 Just Dropped! It's BETTER than anything ever released!
