Raphi-2Code

fan ever

[Raphi-2Code @R2Cdev_]

Three AI agents (Grok 4.7, Claude Opus 5.5, GPT-6 Astra), one Minecraft clone, one goal: Create a harness and then beat the Ender Dragon from a fresh world in under 50h. All three were limited to player inputs only, and none was allowed to modify game files. This was by far the most resource-intensive and expensive benchmark I’ve run so far.

Grok 4.7: Dragon defeated ✅
• 3.1h in-game (~4.1h real time)
• 15 blaze rods, 16 Eyes of Ender, full diamond gear
• Found the stronghold by throwing 2 eyes, destroyed all 10 crystals
• Gray areas it disclosed itself: one portal frame and one test eye throw went through direct function calls instead of simulated clicks, and it used a debug chunk scan and prior knowledge of a chest location
• the model died 36 times in the Minecraft clone

Claude Opus 5.5: Dragon not reached ❌
• After 50h: 2 blaze rods, 5 ender pearls, mixed iron/diamond/gold armor, 14 deaths
• Strict line-of-sight rules, and it found 5 game bugs without exploiting any of them (invisible blazes, an item dupe, a mob it couldn't hit, and more)

GPT-6 Astra: Dragon not reached ❌

• Over 50h of recorded session time, including planning/idle time; exact in-game time wasn’t measured.
• Last inventory: 7 blaze rods, 2 Eyes of Ender, 0 remaining ender pearls; diamond sword and pickaxe, no equipped armor.
• Repeated deaths.
• Used screenshot-based controls, OCR combat/auto-healing helpers and multiple subagents. Stronghold coordinates came from the public README - not from throwing Eyes of Ender.
• Combat and pearl collection stalled progress. Recorder failures caused documented recording gaps.

So currently the leaderboard is:
1) Grok 4.7 (defeated the game)
2) GPT-6 Astra (7 Blaze Rods + 2 Eyes of Ender)
3) Claude Opus 5.5 (2 Blaze Rods + 5 Ender Pearls)

Here are the screenshots of the last inventories for Grok 4.7, GPT-6 Astra and Claude Opus 5.5 and a few evaluation charts:

[Raphi-2Code @R2Cdev_]

Kimi K3.1 Pro lands on https://t.co/q9ne5LAKCY at $1.06/task.
Interesting how most models get worse at chess on my benchmark at the moment, not better.

[Raphi-2Code @R2Cdev_]

@saintXsol I did. Now the limits are way too low.
But back in GPT-5.5 Pro times, the GPT Pro model was my daily driver, even for normal conversations.

[Raphi-2Code @R2Cdev_]

What model is better?

[Raphi-2Code @R2Cdev_]

@adonis_singh yeah faaaar from sota

[Raphi-2Code @R2Cdev_]

@StatsWire teamfood?

[Stats Wire @StatsWire]

🚨 Gemini 4 Carbon (`gemini-4-carbon`) has officially entered Google internal teamfooding.

New identifiers added in the drop:
• gemini-4-carbon
• gemini-4-carbon-teamfood
• carbon-g-medium-tf
• gemini-4-argon-teamfood
• gemini-4-argon-max-mode-tf

Carbon + Argon Max Mode are both live internally 👀

[Raphi-2Code @R2Cdev_]

Claude Opus 5.5 (Max) ranks #1 across every capability category in the Artificial Analysis Capability Index.

[Raphi-2Code @R2Cdev_]

@diegocabezas01 Crazy result but the Suno music still sounds way too sloppy

[Raphi-2Code @R2Cdev_]

@diegocabezas01 🔥

[Raphi-2Code @R2Cdev_]

Claude on the iPhone Duo 👀🤯

[Raphi-2Code @R2Cdev_]

That was six months ago btw.

[6 Months Ago In AI @6monthsAgoInAI]

GPT-5.5 released

https://t.co/9pe7A1K1Cj

[Raphi-2Code @R2Cdev_]

Make it make sense:
Kimi K2.8 was released on September 11, 2026, while Kimi K3 was launched on July 16, 2026.

[Raphi-2Code @R2Cdev_]

@AdamHoltererer More expensive on the AAI cost per task btw

[Raphi-2Code @R2Cdev_]

@DrBeavisAI still prefer the old one id say

[Raphi-2Code @R2Cdev_]

The result after the GPT-6 Pro change looks different, but unfortunately worse.

[Raphi-2Code @R2Cdev_]

I still don’t get subagents btw (also, the result looks very similar to the result 2 weeks ago for my cat SVG test “Create a detailed SVG illustration of a cat.

[Raphi-2Code @R2Cdev_]

FINALLYYYYYY

[Lumina @LuminaBench]

OpenAI have resumed training of their frontier models

This was a lot earlier than I thought so we should be getting something big relatively soon hopefully

[Raphi-2Code @R2Cdev_]

@atomic_chat_hq @cain151714 @Kimi_Moonshot 👀

[Raphi-2Code @R2Cdev_]

@eliancodex Saturday/Sunday

[Raphi-2Code @R2Cdev_]

This has changed very fast! I forgot to add Claude Opus 5 (Max) btw

[Raphi-2Code @R2Cdev_]

Not a single open-weight or Chinese model makes the top 10 of the Artificial Analysis Intelligence Index.

The highest-ranked Chinese model, MiMo-V2.6-Pro, narrowly misses out with a score of 46.3.

[Raphi-2Code @R2Cdev_]

@AdamHoltererer 🔥

[Raphi-2Code @R2Cdev_]

Kimi stole your idea.

[Adam Holter @AdamHoltererer]

GPT distillation detected