New on Ropewalk This Week: Claude Opus 5, Gemini 3.6 Flash, Gemini 3.5 Flash-Lite & Hunyuan Image 3 (July 2026)

Four new models landed on Ropewalk this week: Anthropic's newest Opus-tier flagship, two fresh tiers of Google's Gemini Flash family, and Tencent's first Hunyuan image model on the platform (Hunyuan Video has been live for a while, but this is the family's debut on the image side). All four are still admin-only in the catalog as this article goes to review — they go fully public the instant it's approved and published, so every price and capability below is what you'll actually get when you click through. Three are text models spanning a wide budget range (from a $25/Mtok Opus tier down to a $2.50/Mtok Flash-Lite tier), and the fourth joins an image lineup that already includes ByteDance's Seedream line and Alibaba's Qwen-Image-2 Pro.

By Ropewalk Team. Tested on 2026-07-30 with live generations across all four models — real text replies from Claude Opus 5, Gemini 3.6 Flash, and Gemini 3.5 Flash-Lite, and a real rendered image from Hunyuan Image 3.

What shipped this week — 4 models, July 2026:

• Claude Opus 5 — Anthropic's newest Opus-tier flagship, same $5/$25-per-Mtok pricing tier as Opus 4.8, 200K-token context
• Gemini 3.6 Flash — Google's current Flash-tier model, GA July 2026, faster throughput than 3.5 Flash, 1M-token context
• Gemini 3.5 Flash-Lite — Google's fastest, cheapest current Flash tier at ~350 tokens/second, also 1M-token context
• Hunyuan Image 3 — Tencent's native multimodal image model, the platform's first Hunyuan-family entry on the image side, 320 gems per generation

Between them, the four models cover three very different budgets and two very different media types — a useful spread if you're trying to figure out which tier actually fits a specific task instead of defaulting to whatever's already pinned in your chat.

Claude Opus 5

Claude Opus 5 is Anthropic's newest Opus-tier flagship, released July 2026 as the direct successor to Opus 4.8, and it lands at exactly the same pricing tier: $5 per million input tokens and $25 per million output tokens, billed at 60 gems for a default message. It's built for long-horizon agentic work, deep knowledge tasks, and clear, warm writing rather than raw benchmark chasing, and on Ropewalk it supports a 200,000-token context window with responses tunable up to 4,000 tokens and a dedicated reasoning-token budget of up to 16,000. We verified it with a two-sentence founder-advice prompt: it returned a complete, coherent answer in a single pass at 848 input tokens and 141 output tokens, billing 35 gems for the exchange — confirming the model resolves and prices correctly end to end.

Gemini 3.6 Flash

Gemini 3.6 Flash is Google's current Flash-tier model, reaching general availability in July 2026 as the direct successor to Gemini 3.5 Flash, with faster throughput and the same agentic-and-coding focus. Pricing sits at $1.50 per million input tokens and $7.50 per million output tokens — cheaper on the output side than Gemini 3.5 Flash's $9/Mtok — and Ropewalk exposes its full 1,000,000-token context window alongside temperature, top-P, and top-K controls for advanced users. We tested it with a context-window explainer prompt aimed at coding agents: it returned a real 68-token answer from a 515-token input in a single pass, billing 10 gems for the exchange, with no errors on the Google provider path that has occasionally tripped up other Gemini models on the platform.

Gemini 3.5 Flash-Lite

Gemini 3.5 Flash-Lite is Google's fastest and cheapest current Flash tier, generally available since July 2026 and built for high-volume, low-latency text tasks rather than complex reasoning. It runs at roughly 350 tokens per second, prices at just $0.30 per million input tokens and $2.50 per million output tokens, and is the cheapest of the four models added this week at only 5 gems for a default message. We verified it with a support-automation prompt asking for three quick customer-service use cases: it returned a complete 203-token answer from a 517-token input, billing 5 gems for the exchange — the same gem cost as Kimi K2.6's cheapest tier from last week's roundup, at a fraction of the per-token price.

Hunyuan Image 3

Hunyuan Image 3 is Tencent's native multimodal image model, distributed on Ropewalk as a PrunaAI-squeezed build for faster inference, and it's the first Hunyuan-family model on the image side of the catalog (Tencent's Hunyuan Video model has been live separately for a while). Generation costs a flat 320 gems per image (about $0.08 at the provider level), supports eleven aspect ratios from square through ultra-wide 21:9 and tall 9:21, and defaults to PNG output at 95% quality with a "Go Fast" optimization toggle already enabled by default. We generated a glowing-origami-crane hero image as a verification test: it completed in 19.8 seconds and billed exactly 320 gems, matching the listed rate — and that same real generation is this article's cover image below, not a stand-in from a different model.

New Models at a Glance

Model	Type	Price	Context / Length	Best for
Claude Opus 5	Text	$5 / $25 per Mtok	200K tokens	Long-horizon agentic work
Gemini 3.6 Flash	Text	$1.50 / $7.50 per Mtok	1M tokens	Fast agentic tasks + coding
Gemini 3.5 Flash-Lite	Text	$0.30 / $2.50 per Mtok	1M tokens	High-volume, low-latency text
Hunyuan Image 3	Image	320 gems/image (~$0.08)	11 aspect ratios	Native multimodal image generation

Try them now

All four models are live on Ropewalk as of this week — no waitlist, no separate signup. Claude Opus 5 is the pick when a task genuinely needs Opus-tier depth: long documents, multi-step agentic chains, or writing that has to sound considered rather than fast, and it costs the same 60 gems per message as the Opus 4.8 tier it replaces. Between the two Gemini tiers, Gemini 3.6 Flash is the better default for anything agentic or code-adjacent thanks to its million-token context window and cheaper output pricing, while Gemini 3.5 Flash-Lite is the one to reach for when you're running thousands of short, simple completions and the per-token price matters more than depth — at $0.30/$2.50 per Mtok it's roughly five times cheaper on input than Gemini 3.6 Flash. Hunyuan Image 3 fills a gap the catalog didn't have before — it's the first native multimodal image model from a major Chinese lab available on the platform, and at 320 gems a generation it sits between Nano Banana 2 (160 gems) and GPT Image 2 (500 gems) in the platform's image-pricing ladder — pricier than budget tiers like Seedream 4 (25 gems) but well under the top end.

Every model here passed a real end-to-end generation before shipping in this article: three real text replies with token counts and gem costs that matched their listed pricing exactly, and one real rendered image that billed the same 320 gems quoted above. That's the same bar the four models in last week's roundup had to clear.

If you're deciding between Ropewalk's text models more broadly, see our ChatGPT vs Claude vs Gemini comparison. For image models, our four-way head-to-head of GPT Image 2, Nano Banana 2, Imagen 4 Ultra, and FLUX 2 Pro shows how the existing lineup compares.