New LLMs, Oct 2026: Gemini 3.5 Flash Lite, Nano Banana 2.1

This is the October 2026 sibling of the September 2026 new-model guide, written the same way: for each catalog or plan change recorded between that post's wrap-up (as of 2026-09-30) and 2026-10-09, what the live catalog and the plan responses show for the ids involved — which endpoints serve them, whether a plan covers them — and how to send a first request. As of the 2026-10-09 catalog, Router One no longer lists Claude Fable 5.1 (anthropic/claude-fable-5.1) and lists two new ids, Gemini 3.5 Flash Lite (google/gemini-3.5-flash-lite) and Gemini Nano Banana 2.1 (gemini-nano-banana-2.1). On the plan side, GPT-6.1 Sol and GPT-6 Sol are in the Premium models tier of the Pro, Max and Ultra plans, first seen in the 2026-10-06 plan response.

The 2026-10-09 catalog holds 67 ids — 59 text and 8 image. No plan tier in the 2026-10-09 plan response lists Gemini 3.5 Flash Lite or Gemini Nano Banana 2.1, so their calls bill from the wallet balance at their posted rates; Gemini 3.5 Flash stays in the Standard models tier. Listing dates in this post are catalog observation dates, not vendor launch dates, and the changelog records the same changes.

October at a glance (as of 2026-10-09)

The table lists each change, the endpoints that serve each id and whether a plan covers it; live rates are on each model page, not here.

Claude Fable 5 (anthropic/claude-fable-5) stays listed. Plan model lists change, and the pricing page shows the live ones with each plan's allowance; a model that no plan tier lists bills from the wallet, per token or per image, whether or not you hold a plan.

GPT-6.1 Sol and GPT-6 Sol in the Premium models tier

As of the 2026-10-06 plan response, GPT-6.1 Sol (openai/gpt-6.1-sol) and GPT-6 Sol (openai/gpt-6-sol) are in the Premium models tier of the Pro, Max and Ultra plans, beside GPT-5.6 Sol, GPT-5.5, Claude Opus 5 and the other models that tier lists; the 2026-10-03 plan response listed neither. On a plan, their calls count against the Premium models allowance — 1 request per call, 2 when the total input is strictly above 272,000 tokens — and calls beyond the allowance bill the wallet at the posted rates; the pricing page lists each plan's allowance.

Both are GPT-family ids, served on Chat Completions and natively on Responses, the wire format Codex CLI speaks. On a Pro, Max or Ultra plan, GPT-6.1 Sol counts against the Premium models allowance (as of the 2026-10-06 plan response), and moving Codex from gpt-5.6-sol to gpt-6.1-sol keeps it on the same allowance. Without a plan, every request Codex sends bills the wallet at GPT-6.1 Sol's posted rates (model page). The GPT-6.1 Sol API guide and the GPT-6 Sol API guide cover both ids, both endpoints and Codex CLI setup.

Gemini 3.5 Flash Lite

Router One lists Gemini 3.5 Flash Lite as google/gemini-3.5-flash-lite as of the 2026-10-09 catalog. Google writes the name "Gemini 3.5 Flash-Lite"; it made the model generally available on July 21, 2026 and describes it as a low-latency option designed for high-volume automation, per the Gemini API release notes (checked 2026-10-09). The catalog entry lists a 1,048,576-token context window, text and image input and the chat, streaming, tool-calling and vision flags; as a Gemini id it is served on POST /v1/chat/completions. It is in no plan tier as of the 2026-10-09 plan response, so its calls bill from the wallet at the posted rates on the Gemini 3.5 Flash Lite model page, plan or no plan.

Gemini 3.5 Flash (google/gemini-3.5-flash) is a different model: as of the 2026-10-09 plan response it is in the Standard models tier of the Pro, Max and Ultra plans. Gemini 3.1 Flash Lite (google/gemini-3.1-flash-lite, with its preview twin google/gemini-3.1-flash-lite-preview) is an earlier Flash Lite generation, also in no plan tier in the same response. Per Google's guide to Gemini 3.6 Flash and 3.5 Flash-Lite (checked 2026-10-09), starting with those two models the Gemini API ignores temperature, top_p and top_k and rejects a request whose last turn is a prefilled model turn, and Gemini 3.5 Flash Lite's default thinking level is minimal; Gemini 3.1 Flash Lite and Gemini 3.5 Flash predate these changes. Move tone and format rules into the system message before you switch, then compare on your own prompts: Gemini 3.5 Flash Lite vs Gemini 3.1 Flash Lite and Gemini 3.5 Flash Lite vs Gemini 3.5 Flash render the spec sheets and live rates side by side.

Access from mainland China for the whole Gemini family is on the Gemini API in China page.

Gemini Nano Banana 2.1

Google released Gemini Nano Banana 2.1 on October 6, 2026 as an update to Nano Banana 2, per the Gemini API release notes (checked 2026-10-09), and its image generation guide (checked 2026-10-09) calls 2.1 the primary high-efficiency model for image generation and conversational editing and recommends it over Nano Banana 2 for new projects. Router One lists it as gemini-nano-banana-2.1 as of the 2026-10-09 catalog: text and image input, image output, billed per generated image — its live unit price is on the Gemini Nano Banana 2.1 model page. No plan tier in the 2026-10-09 plan response lists it, so each image bills from the wallet balance.

It takes the same two calls as Nano Banana 2 (gemini-3.1-flash-image-preview) and Nano Banana Pro (gemini-3-pro-image-preview): put gemini-nano-banana-2.1 in model on POST /v1/images/generations, or send a reference image to POST /v1/images/edits. The size and quality notes in the Nano Banana 2 API guide were written for Nano Banana 2; Google's model page (checked 2026-10-09) lists 1K, 2K and 4K output for 2.1 and its release notes describe improved wide and panoramic aspect-ratio generation, so describe the framing you want in the prompt and check the result. The image generation API page covers the image models on the gateway, each with its own per-image price on its model page.

Claude Fable 5.1 and the Claude Code aliases

As of the 2026-10-09 catalog, Router One no longer lists Claude Fable 5.1 (anthropic/claude-fable-5.1). Claude Fable 5 (anthropic/claude-fable-5) stays listed and, as a Claude-family id, is served on POST /v1/messages and POST /v1/chat/completions, not on POST /v1/responses. The September post and the changelog keep their September entries as dated history.

In Claude Code this shows up through the fable alias: pin it, and keep the Haiku line too if you use the haiku alias. Add these lines to your shell profile (~/.zshrc or ~/.bashrc):

export ANTHROPIC_DEFAULT_HAIKU_MODEL=claude-haiku-4-5
export ANTHROPIC_DEFAULT_FABLE_MODEL=claude-fable-5

If the one-click install wrote your setup, merge the same keys into the env block you already have in ~/.claude/settings.json instead:

{
  "env": {
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "claude-haiku-4-5",
    "ANTHROPIC_DEFAULT_FABLE_MODEL": "claude-fable-5"
  }
}

The Haiku line is optional: it runs background tasks on Claude Haiku 4.5 rather than the main model. Per Claude Code's changelog, model configuration and subagents docs (Claude Code 2.1.295, checked 2026-10-09), from 2.1.293 the haiku alias resolves to Claude Haiku 5.5 (claude-haiku-5-5), which the 2026-10-09 Router One catalog does not include, so keep the line if you use /model haiku, subagents defined with model: haiku or the built-in claude-code-guide subagent (which the subagents docs list as running on Haiku). The Fable line points the fable alias at Claude Fable 5: per Claude Code's model configuration docs (Claude Code 2.1.295, checked 2026-10-09), the alias otherwise resolves to Claude Fable 5.1 (claude-fable-5-1), which Router One no longer lists as of the 2026-10-09 catalog. Claude Fable 5 is in no plan tier as of the 2026-10-09 plan response, so it bills to the wallet.

Plan holders also pin the sonnet and opus aliases and the main model; the Claude Code setup guide has the full block and explains each line.

Which endpoint serves which id

Endpoint families are not interchangeable. POST /v1/chat/completions serves every chat model in the catalog; POST /v1/messages serves the Claude-family and DeepSeek ids; POST /v1/responses is served natively for the GPT-family and DeepSeek ids and for the Grok chat ids. A Claude-family id sent to /v1/responses, or a chat id outside the Claude and DeepSeek families sent to /v1/messages, is rejected with HTTP 400 invalid_request_error before any model is called, and the message names the path to use. Image models take the Images API instead: POST /v1/images/generations for a prompt, POST /v1/images/edits for a prompt plus a reference image. For the ids in this post that means: GPT-6.1 Sol and GPT-6 Sol on Responses or Chat Completions; Gemini 3.5 Flash Lite on Chat Completions; Gemini Nano Banana 2.1 on the two image endpoints; and Claude Fable 5 on Messages or Chat Completions. The OpenAI-compatible API page covers the request shape, and each model page lists exactly the endpoints it serves.

How to try one in five minutes

1. Create a key. Dashboard → API Keys → Create Key. Keys look like sk-.... If this is an experiment, give the key a maxSpend cap so a runaway loop stops at a number you chose (per-key cost tracking).
2. Set the base URL. https://api.router.one/v1 for OpenAI-compatible clients and SDKs; the chat and image endpoints both sit under it and take the same key.
3. Send one request per model.

Gemini 3.5 Flash Lite on Chat Completions:

curl https://api.router.one/v1/chat/completions \
  -H "Authorization: Bearer sk-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "google/gemini-3.5-flash-lite",
    "messages": [
      {"role": "system", "content": "Classify the ticket as billing, bug or question. Reply with one word."},
      {"role": "user", "content": "The invoice shows the wrong company name."}
    ]
  }'

Gemini Nano Banana 2.1 on the Images API:

curl https://api.router.one/v1/images/generations \
  -H "Authorization: Bearer sk-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-nano-banana-2.1",
    "prompt": "Isometric illustration of a night market, warm lantern light, no text"
  }'

Every image the call returns is billed at the per-image unit price on the model page. The Nano Banana 2 API guide shows how to handle the response and how to send a reference image to /v1/images/edits, and the Images API docs list the fields.

4. Read the trace. Dashboard → Logs lists both calls with the model, cost, latency and status, so you see what each call cost before you wire either model into your app (per-request observability).

Top up with a card or Alipay through one hosted checkout, or with USDT/USDC on six chains (Tron, BSC, Ethereum, Polygon, Base, Arbitrum). No US credit card required. Requests reach the catalog from mainland China without a VPN.

FAQ

What is the model id for Gemini 3.5 Flash Lite, and does a plan cover it? The id is google/gemini-3.5-flash-lite, called on /v1/chat/completions with a Router One key. It is in no plan tier as of the 2026-10-09 plan response, so its calls bill from the wallet balance at the posted rates on its model page, plan or no plan. Gemini 3.5 Flash (google/gemini-3.5-flash) is a different model, and the Standard models tier of the Pro, Max and Ultra plans lists it in the same response.

Should I use Gemini 3.5 Flash Lite or Gemini 3.1 Flash Lite? Both are Chat Completions ids with a 1,048,576-token context window and text and image input in the 2026-10-09 catalog, and both are in no plan tier as of the 2026-10-09 plan response. Per Google's guide to Gemini 3.6 Flash and 3.5 Flash-Lite (checked 2026-10-09), starting with those two models the Gemini API ignores temperature, top_p and top_k and rejects a request whose last turn is a prefilled model turn, and Gemini 3.5 Flash Lite's default thinking level is minimal; Gemini 3.1 Flash Lite predates these changes. Move tone and format rules into the system message, run your own prompts through both and compare each request's cost in Dashboard → Logs; Gemini 3.5 Flash Lite vs Gemini 3.1 Flash Lite renders both spec sheets and live rates.

Is Gemini 3.5 Flash Lite the same model as Gemini 3.5 Flash? No. google/gemini-3.5-flash-lite and google/gemini-3.5-flash are two catalog ids with their own rates. As of the 2026-10-09 plan response, Gemini 3.5 Flash is in the Standard models tier of the Pro, Max and Ultra plans and Gemini 3.5 Flash Lite is in no plan tier. On a plan, the choice therefore also decides whether a call draws plan quota or bills the wallet. Per Google's guide to Gemini 3.6 Flash and 3.5 Flash-Lite (checked 2026-10-09), Gemini 3.5 Flash predates the sampling-parameter and prefill changes that start with those two models; Gemini 3.5 Flash Lite vs Gemini 3.5 Flash puts the two side by side.

Should I use Nano Banana 2.1, Nano Banana 2 or Nano Banana Pro? All three are in the 2026-10-09 catalog — gemini-nano-banana-2.1, gemini-3.1-flash-image-preview and gemini-3-pro-image-preview — and take the same Images API requests, each billed per image at the unit price on its own model page. Google's image generation guide (checked 2026-10-09) describes Nano Banana 2.1 as an update to Nano Banana 2 and the primary high-efficiency model for image generation and conversational editing, recommends it over Nano Banana 2 for new projects, and positions Nano Banana Pro for the most complex visual tasks. Run your own prompts through each: swap the model string, compare the images, and compare each request's cost in Dashboard → Logs.

Do GPT-6.1 Sol and GPT-6 Sol count against a plan's allowance? Yes, on a Pro, Max or Ultra plan. Both are in the Premium models tier, first seen in the 2026-10-06 plan response, so each call draws the Premium models allowance — 1 request per call, 2 when the total input is strictly above 272,000 tokens — and calls beyond the allowance bill the wallet at the posted rates. Without a plan, their calls bill per token from the wallet. The pricing page lists each plan's allowance.

Can I still use Claude Fable 5.1? As of the 2026-10-09 catalog, Router One no longer lists Claude Fable 5.1, so call Claude Fable 5 (anthropic/claude-fable-5) by id instead. In Claude Code, pin the fable alias with ANTHROPIC_DEFAULT_FABLE_MODEL=claude-fable-5 (the fable alias otherwise requests claude-fable-5-1, which Router One no longer lists as of the 2026-10-09 catalog). Claude Fable 5 is in no plan tier as of the 2026-10-09 plan response, so it bills to the wallet.

Next steps

• Check the Gemini 3.5 Flash Lite and Gemini Nano Banana 2.1 model pages for live rates and capability flags.
• Compare before you switch: Gemini 3.5 Flash Lite vs Gemini 3.1 Flash Lite and Gemini 3.5 Flash Lite vs Gemini 3.5 Flash.
• Generate and edit images with the Nano Banana 2 API guide and the image generation API page.
• See which models each plan tier lists, and each plan's allowance, on the pricing page.
• Read the September 2026 guide for the previous batch — GPT-6 Sol, Claude Opus 5.5, GPT-6 Astra and GPT-6.1 Sol among them.
• Set up Claude Code with the Claude Code setup guide, or point Codex CLI at a GPT id with the Codex and Responses API guide.