A New Vision Leader and Fresh Contenders
For the first time, claude-fable-5 has claimed the top spot on the Vision leaderboard, overtaking claude-opus-4-7-thinking — marking a significant Anthropic-on-Anthropic shakeup in a category long dominated by Google’s Gemini family. Meanwhile, two new models have broken into the top 10 across Vision and Text-to-Image, signaling that the mid-2026 model race is heating up fast.
What Changed Today
- Vision — New #1: claude-fable-5 overtook claude-opus-4-7-thinking to become the new leader in Vision.
- Vision — New Top 10 Entry: gpt-5.5 entered the Vision top 10 with an ELO of 1285.
- Text-to-Image — New Top 10 Entry: qwen-image-2.0-pro-2026-06-22 entered the Text-to-Image top 10 with an ELO of 1193.
Vision: Claude Fable 5 Claims the Crown
claude-fable-5 is now the #1 Vision model, ending a stretch where Google’s Gemini 3 Pro and its variants dominated the top of this leaderboard. This is notable because it’s the first time an Anthropic model has held the outright Vision lead — and the “Fable” naming suggests this may be a new model family entirely, not just an incremental Opus or Sonnet update. Developers building vision-heavy pipelines (document understanding, image analysis, multimodal agents) should start benchmarking against this model immediately. Pricing and API model ID have not yet been confirmed in public documentation, so watch Anthropic’s API changelog closely.
Also worth flagging: gpt-5.5 from OpenAI entered the Vision top 10 at an ELO of 1285 — just barely behind where Gemini 3 Pro has been sitting at 1288. This puts three providers (Anthropic, Google, OpenAI) within striking distance of each other at the top of Vision for the first time in months. If you’re locked into the OpenAI ecosystem, gpt-5.5 appears to be a meaningful step up from the gpt-5.1 and gpt-5.2 variants already in the top 10. Keep an eye out for the official API model ID — likely gpt-5.5 — and pricing details from OpenAI.
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: Alibaba’s Qwen Breaks Into the Top 10
qwen-image-2.0-pro-2026-06-22 has entered the Text-to-Image top 10 with an ELO of 1193, slotting in above Tencent’s Hunyuan Image 3.0 (1151) and close to Google’s Gemini 2.5 Flash image preview (1156). This is a significant entry for Alibaba’s Qwen team, which has been steadily climbing across multiple modalities this year. For developers looking for a strong non-Western-provider option for image generation — particularly if you’re serving markets where Alibaba Cloud infrastructure is advantageous — this is worth testing. The API model ID is likely qwen-image-2.0-pro-2026-06-22 and pricing should be available through Alibaba Cloud’s Model Studio. The overall leader remains OpenAI’s gpt-image-1.5-high-fidelity (ELO: 1247, API ID: gpt-image-1) at $0.02–$0.19 per image.
Current Leaders at a Glance
| Category | #1 Model | Provider | Score (ELO) |
|---|---|---|---|
| Vision | claude-fable-5 | Anthropic | New leader (exact ELO TBC) |
| Text-to-Image | gpt-image-1.5-high-fidelity | OpenAI | 1247 |
So What?
Today’s biggest takeaway: the Vision category just got genuinely competitive across three major providers. If you’ve been defaulting to Gemini 3 Pro for vision tasks, now is the time to run structured evals against claude-fable-5 and gpt-5.5 — the ELO gaps are razor-thin, and the right choice for your use case may come down to pricing, latency, or context window rather than raw quality. For Text-to-Image builders, qwen-image-2.0-pro is a credible new option worth benchmarking, especially if you need provider diversification or are optimizing for cost. As always, don’t swap models in production based on leaderboard moves alone — run your own evals on your own data, but today’s shifts tell you exactly which models deserve a spot in your next evaluation cycle.
