Today’s Ranking Shakeup: Two New Contenders Enter the Arena
Two fresh models cracked the top 10 today across code generation and text-to-image — Anthropic’s Claude Opus 4.7 Thinking lands just below the reigning code champion, while a mysterious newcomer called MAI Image 2.6 Preview muscles its way into the image generation leaderboard with an impressive debut score. Here’s what builders need to know.
What Changed Today (August 11, 2026)
- Code Generation: claude-opus-4-7-thinking is new in the top 10 with an ELO of 1558 — landing at near the very top of the leaderboard, just 3 points behind the current leader.
- Text-to-Image: mai-image-2.6-preview is new in the top 10 with an ELO of 1336 — a significant debut that would place it well above every existing model on the leaderboard.
Code Generation: Claude Opus 4.7 Thinking Nips at the Crown
Anthropic’s Claude Opus 4.7 Thinking has entered the code generation rankings with an ELO of 1558, placing it just 3 points behind the current leader, Claude Opus 4.6 (ELO: 1561). This is the thinking-enabled variant of what appears to be Anthropic’s next-generation Opus release, and the fact that it debuts this close to the top suggests it could overtake the leader as more votes come in. For context, the previous thinking variant (claude-opus-4-6-thinking) sits at 1551, so the 4.7 thinking model represents a meaningful jump in deep-reasoning code performance.
For developers, this is worth watching closely. If you’re already using claude-opus-4-6 with thinking enabled for complex coding tasks — agentic workflows, multi-file refactors, or architectural reasoning — keep an eye on claude-opus-4-7-thinking as the likely API model ID. Pricing hasn’t been confirmed yet, but Anthropic’s Opus tier has historically sat at $5/$25 per MTok (input/output). The current code leaderboard leader remains Claude Opus 4.6, but this new model is knocking on the door.
FREE GUIDE
Stop Writing Design Specs by Hand
Get the free visual guide: how AI tools generate GAMP 5 documentation directly from your PLC and DCS exports. Used by Life Sciences engineers who are done doing it manually.
No spam. Unsubscribe anytime.
Text-to-Image: MAI Image 2.6 Preview Makes a Stunning Entrance
mai-image-2.6-preview has debuted in the text-to-image rankings with an ELO of 1336 — and this is a big deal. That score doesn’t just place it in the top 10; it would put it nearly 90 ELO points above the current leader, OpenAI’s gpt-image-1.5-high-fidelity (ELO: 1247). If this holds as more votes accumulate, we could be looking at a new category leader from an unexpected provider. The “MAI” branding and preview tag suggest this is either a new entrant or a rebranded effort — details on the provider, pricing, and API access are still emerging.
Developers building image generation pipelines should take note but proceed with caution. Preview models can shift significantly in score as vote counts grow (the current leaders have tens of thousands of votes baking in their ELOs). That said, an 89-point ELO gap is not a rounding error — it signals genuinely superior output quality in head-to-head comparisons. Watch for mai-image-2.6-preview API availability and pricing details. If you’re currently using gpt-image-1 or gemini-3-pro-image-preview, this is worth benchmarking against your specific use cases immediately once access opens up.
Current Leaders at a Glance
| Category | #1 Model | Provider | Score (ELO) |
|---|---|---|---|
| Code Generation | claude-opus-4-6 | Anthropic | 1561 |
| Text-to-Image | gpt-image-1.5-high-fidelity* | OpenAI | 1247 |
*Note: mai-image-2.6-preview has debuted at ELO 1336, which would surpass the current leader. We’re tracking this as a potential leader change pending further vote stabilization.
So What?
If you’re building with code generation APIs, the practical move today is straightforward: Claude Opus 4.6 remains your best bet, but start testing claude-opus-4-7-thinking for reasoning-heavy coding tasks the moment it’s available in your tier — a new Opus generation closing to within 3 ELO points on arrival strongly suggests it will take the top spot once votes mature. For image generation, mai-image-2.6-preview is the more disruptive story. An 89-point ELO gap over GPT-Image-1.5 is enormous if it holds, and could reshape which API you default to for production image workflows. The smart play: get on any waitlist or preview program for MAI Image 2.6 now, run your own evals against your actual prompts, and be ready to switch if quality holds at scale. Early movers who integrate the best models fastest ship better products — that’s the whole game.
