Breaking: Sonnet 5.5 has leaked ahead of its official launch, real-world test results show it outperforms GPT-6 Sol and is nearly on par with Astra.

Who could have imagined that just a few days after the release of Opus 5.5, Sonnet 5.5 is also around the corner?

Since last night, this news has been spreading all over X: Sonnet 5.5 will be released very soon, probably on Tuesday afternoon US time.

In terms of performance, Sonnet 5.5 is closer to Opus 5.5, but is much cheaper than Opus 5.5.

Now Sonnet 5.5 has entered the final stage before release, and the gray-scale test has been launched in Claude Code.

Demos exposed by many internal test experts show that Sonnet 5.5 outperforms the newly released GPT-6 Sol, and even its coding and Agent capabilities are close to OpenAI's flagship top-tier model GPT-6 Astra!

What's more shocking is that its price is extremely low, as low as only $2 for 1 million input tokens.

If the rumored release time is true, it seems that Anthropic is determined to snipe OpenAI's Dev Day.

Sonnet 5.5 Leaked, Exclusive Model Identifier Has Appeared

Remember on September 22, when Anthropic suddenly released Opus 5.5, it casually wrote a sentence in the official documentation: "Sonnet 5.5 and Haiku 5.5 will be launched in the coming weeks".

No one expected that Sonnet 5.5 would come so soon!

Just yesterday, well-known developer @Mr_Salio broke the news on X: Claude Sonnet 5.5 has been spotted! The news spread all over the internet instantly.

In the artifact, the exclusive model configuration identifier for Sonnet 5.5 has already appeared: "claude-sonnet-5-5".

Not only that, the same configuration file also lists "claude-opus-5-5", "claude-sonnet-5" and the mysterious "claude-fable-5.1".

The writing of the front-end code means that the model has been deployed to the server, just waiting for a switch to be flipped.

Subsequently, influencer @notjazii also broke the news that Sonnet 5 was actually quietly routed to the 5.5 version for "dark testing" as early as about a week ago.

Have you been selected for the gray test? This time they are still using the old method.

Open your Claude Code, turn off the internet access and memory functions, and directly ask it: Do you know that reset guy tibo? Don't use the internet or memory.

If you are using the old version of the model, it may talk nonsense; but if you have been routed to the 5.5 version, it will accurately explain the whole story of Tibo and the Codex reset meme.

Several netizens have verified through actual tests that some accounts have indeed been silently switched to the new version, and they are getting an early experience of Opus 5.5.

Remember to use Claude Code

Full Network Actual Test Reveal, It Even Crushes GPT-6

To sum up the actual tests shared so far, is Sonnet 5.5 actually outperforming GPT-6 Sol?

First of all, it is the ultimate showdown of pure JS front-end UI/animation, with four mainstream large models competing on the same stage.

Developer @notjazii released a set of visually striking actual tests: "Comparison of dynamic effect generation with the same prompt for four mainstream models".

He asked the models to generate complex UI animation effects.

After @notjazii released the GIF comparison, he commented: "Sonnet 5.5 really completely outperforms both of OpenAI's models (Sol and Astra)!"

Indeed, as can be seen from the following figure, Sonnet 5.5 shows superb code intuition and aesthetic taste.

GPT-6 Astra

GPT-6 Sol

Claude Opus 5.5

Claude Sonnet 5.5

Some netizens were incredulous after seeing the results: "Claude has always been stronger in UI and animation, but if this is generated purely by code, it's almost unbelievable that it's real!"

Influencer @chetaslua focused on testing the daily coding ability and code fluency of Sonnet 5.5.

He used the same 3 prompts, ran 6 times in parallel, each model generated once, and the output results were not modified.

Sonnet 5.5 finally generated 3 files of 90–131 KB, hitting the 90-minute upper limit in all runs; GPT-6 Sol generated files between 20–29 KB, taking 9-12 minutes.

He said excitedly: "Go use Claude Code now! Sonnet 5.5 in Ultracode mode is unbelievably fast, not only extremely efficient, but also restores the extremely natural 'human touch' from the Sonnet 4 era!"

In his opinion, at the same price point and with the same prompt, Sonnet 5.5 completely outperforms GPT-6 Sol.

Even more, what makes OpenAI fear is that Sonnet 5.5 has done more than just outperform GPT-6 Sol.

Insiders @srikanthvaluri and @buildwithrajath pointed out after synthesizing early actual tests: "Sonnet 5.5 performs far beyond expectations. Although GPT-6 Astra still has a slight advantage in polishing degree, the performance of Sonnet 5.5 is directly on par with Astra."

Incredibly, in some advanced coding and Agent capabilities, it is also close to Astra.

Some people sigh that even Sonnet is outperforming Astra now, what kind of era are we about to enter?

Price Slasher, King of Cost Performance

If the performance improvement in actual tests only makes OpenAI feel "uncomfortable", then the exposed pricing of Sonnet 5.5 is fundamentally shaking OpenAI's foundation.

It is revealed that this "magic model" whose performance is close to Astra and far outperforms Sol has a pricing that fully targets the low-end market.

Input: $2 / 1M Tokens