8 New Ropewalk Models (Nov 2026): GPT-6.1 Terra & More

By Ropewalk Team. Tested on 2026-11-06: we ran a live 1M-context prompt on GPT-6.1 Terra and read the full reply.

Between 29 October and 5 November 2026, Ropewalk added 8 new text models, and every one of them offers a context window of about 1M tokens. This roundup shows what each model does, what it costs to try in coins, and which one to open first for which job.

Quick answer

Eight models went live on 2026-10-29 to 2026-11-05: GPT-6.1 Terra, GPT-6.1 Terra Pro, Kimi K3.1 Pro, GLM-5.3 Prime, NVIDIA Nemotron 3.5 Super VL, MiniMax M3.1 Flash, Mercury 3 and inclusionAI Ling 3.1 Flash. For a strong all-round model start with GPT-6.1 Terra (). For long coding and agent work pick Kimi K3.1 Pro () or GLM-5.3 Prime (). For quick, cheap chats use Ling 3.1 Flash (), MiniMax M3.1 Flash or Mercury 3 ( each). For photos, charts and video frames, use Nemotron 3.5 Super VL ().

Headliners

GPT-6.1 Terra and Terra Pro: the mid-tier GPT-6.1 with a 1.05M-token window

GPT-6.1 Terra is OpenAI's mid-tier model of the GPT-6.1 series, with a 1.05M-token context window, which is roughly 700,000 words in a single conversation. GPT-6.1 Terra Pro is the same model served in pro reasoning mode for harder tasks. Both start at per message in Ropewalk chat, so switching to the Pro mode for a difficult question does not change the starting price. We ran the prompt below on GPT-6.1 Terra on 2026-11-06 and got three tidy bullets in a few seconds for 10 coins: whole-codebase bug tracing, comparing clauses across 200 contracts, and summarizing a year of meeting transcripts. Use Terra for everyday work and Terra Pro when the answer needs more careful reasoning.

Kimi K3.1 Pro: open-weight reasoning for long coding sessions

Kimi K3.1 Pro is Moonshot AI's open-weight multimodal reasoning model, built for long-horizon coding with a 1M-token context window. It is the most expensive of the eight at , and it earns that price on jobs that run long: reading a large repository, planning a multi-file change and keeping track of it over many steps. Because it is multimodal, you can also attach a screenshot of an error or a design mock-up next to your question. Start with a real task from your own project rather than a toy example, since the long context is the whole point of the model.

GLM-5.3 Prime: Z.ai's top tier for agent work

GLM-5.3 Prime is the top-tier model from Z.ai, with a 1M-token context window, built for long-horizon coding and agent work. At per message it sits just below Kimi K3.1 Pro, so the two make a natural head-to-head on the same task. Choose it when a job has many steps: planning, calling tools, checking results and revising. Paste your task, ask for a plan first and then ask it to carry the plan out step by step.

NVIDIA Nemotron 3.5 Super VL: image and video input at 10 coins

NVIDIA Nemotron 3.5 Super VL is a 120B mixture-of-experts vision-language model with 12B active parameters and a 1M-token context window. It accepts image and video input, which is rare at this price: a message starts at . Attach a chart, a photo of a whiteboard or a few video frames, and ask it to describe, extract or explain what it sees. It is the most affordable way in this batch to point a model at visual material and get a written answer.

Fast and cheap tier

Three of the eight models are built for speed and low cost, and all three keep the 1M-token window. Each card shows the live cost.

MiniMax M3.1 Flash

MiniMax M3.1 Flash is a fast multimodal model with a 1M-token context window, strong at agentic tasks and coding, starting at . It is a good default for quick coding questions and for chats where you attach an image.

Mercury 3

Mercury 3 is Inception's diffusion-based language model, built for low-latency agentic workflows, with a 1M-token context window and a starting price of . Instead of writing one token after another, a diffusion model refines the whole answer in passes, which is why it suits loops where an agent calls the model many times in a row.

inclusionAI Ling 3.1 Flash

inclusionAI Ling 3.1 Flash is a fast hybrid-reasoning mixture-of-experts model with 560B total and 25B active parameters and a 1M-token window. At it is the cheapest of the eight, so use it for drafts, summaries and high-volume questions where speed matters more than depth.

Which model for which job

All 8 models in one table, with the starting price in coins. Pick the job first, then open the card to see the live cost.

Model	Context	Input	Starting price	Best for
GPT-6.1 Terra	1.05M	Text		everyday all-round work
GPT-6.1 Terra Pro	1.05M	Text		harder problems, careful reasoning
Kimi K3.1 Pro	1M	Text, images		long coding sessions
GLM-5.3 Prime	1M	Text		multi-step agent work
NVIDIA Nemotron 3.5 Super VL	1M	Text, images, video		reading images and video
MiniMax M3.1 Flash	1M	Text, images		fast coding and agent tasks
Mercury 3	1M	Text		low-latency agent loops
inclusionAI Ling 3.1 Flash	1M	Text		cheap drafts and summaries

FAQ

How many new models arrived between 29 October and 5 November 2026?

Eight text models: GPT-6.1 Terra, GPT-6.1 Terra Pro, Kimi K3.1 Pro, GLM-5.3 Prime, NVIDIA Nemotron 3.5 Super VL, MiniMax M3.1 Flash, Mercury 3 and inclusionAI Ling 3.1 Flash.

Which new model is the cheapest to try?

inclusionAI Ling 3.1 Flash starts at . MiniMax M3.1 Flash, Mercury 3 and NVIDIA Nemotron 3.5 Super VL start at . The live cost is always shown on each model card.

Which new models can read images or video?

NVIDIA Nemotron 3.5 Super VL accepts image and video input. Kimi K3.1 Pro and MiniMax M3.1 Flash are multimodal, so you can attach images to your message.

What is the difference between GPT-6.1 Terra and Terra Pro?

Both share the 1.05M-token context window. Terra Pro is the same model in pro reasoning mode, meant for harder tasks.

Earlier releases: 7 new models from October 2026, Nano Banana 2.1, FLUX 3 and 13 new models and 18 new models from September 2026. To start with any model above, open Ropewalk and pick it in a new chat.