New AI models
AI model releases as they appear on SF·SIX, grouped by week, with each model’s first-seen date and live API price.
This week (since Monday 2026-10-12, UTC) SF·SIX first listed 8 new AI models: DeepSeek V4.1 Flash 1017, Bytedance Omnihuman V2, FLUX.3 Klein Image Editing, NVIDIA Nemotron 3.5 Super Text, Muse Spark Open and others. In the last 8 weeks: 32 releases from 14 developers, most recently DeepSeek V4.1 Flash 1017 (2026-10-17). Each links to its live prices.
This week (since 2026-10-12)
| Model | Developer | First seen (UTC) | Released | Headline price | Providers |
|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | DeepSeek | 2026-10-17 | 2026-10-17 | $0.10 / $0.50input / output per 1M tokens | 2 |
| Bytedance Omnihuman V2 | ByteDance Seed | 2026-10-16 | 2026-10-15 | $0.105 per second | 1 |
| FLUX.3 Klein Image Editing2 endpoints | Black Forest Labs | 2026-10-16 | 2026-10-14 | $0.009 per megapixel | 1 |
| NVIDIA Nemotron 3.5 Super Text3 endpoints | NVIDIA | 2026-10-16 | 2026-10-15 | $0.01 per 1000 tokens | 1 |
| Muse Spark Open | Meta | 2026-10-15 | 2026-10-15 | $1.25 / $4.25input / output per 1M tokens | 5 |
| Nemotron 3.5 Super VL | NVIDIA | 2026-10-15 | 2026-10-15 | $0.10 / $0.40input / output per 1M tokens | 5 |
| Gemini 4 Argon | Google DeepMind | 2026-10-13 | 2026-10-13 | $2.00 / $10.00input / output per 1M tokens | 2 |
| Muse Wonder | Meta | 2026-10-12 | 2026-10-12 | $4.00 / $20.00input / output per 1M tokens | 1 |
Week of 2026-10-05
| Model | Developer | First seen (UTC) | Released | Headline price | Providers |
|---|---|---|---|---|---|
| Claude Haiku 5.5 | Anthropic | 2026-10-08 | 2026-10-07 | $0.10 / $0.50input / output per 1M tokens | 5 |
| Nano Banana 2.12 endpoints | Google DeepMind | 2026-10-07 | 2026-10-06 | $1.00 per unit | 1 |
| Nano Banana 2.1 | Google DeepMind | 2026-10-06 | 2026-10-06 | $1.50 / $7.50input / output per 1M tokens | 1 |
| Mistral Large 4 | Mistral AI | 2026-10-06 | 2026-10-06 | $0.68 / $2.09input / output per 1M tokens | 1 |
Week of 2026-09-28
| Model | Developer | First seen (UTC) | Released | Headline price | Providers |
|---|---|---|---|---|---|
| Grok Imagine Video 1.5 Lite Text to Video | xAI | 2026-10-02 | 2026-10-01 | $0.01 per unit | 1 |
| GPT-6.1 Sol | OpenAI | 2026-09-29 | 2026-09-29 | $2.00 / $10.00input / output per 1M tokens | 3 |
| GPT-6.1 Sol Pro | OpenAI | 2026-09-29 | 2026-09-29 | $2.00 / $10.00input / output per 1M tokens | 2 |
| Claude Sonnet 5.5 | Anthropic | 2026-09-28 | 2026-09-28 | $2.00 / $10.00input / output per 1M tokens | 5 |
Week of 2026-09-21
| Model | Developer | First seen (UTC) | Released | Headline price | Providers |
|---|---|---|---|---|---|
| Gemini 3.8 Flash Lite TTS | Google DeepMind | 2026-09-27 | 2026-09-25 | $0.03 per 1,000 characters | 1 |
| Gemini 3.8 Flash TTS | Google DeepMind | 2026-09-27 | 2026-09-25 | $0.045 per 1,000 characters | 1 |
| Seedream (flash / edit)3 endpoints | ByteDance Seed | 2026-09-27 | 2026-09-23 | $0.027 per image | 1 |
| GLM 5.3 Prime | Z.ai (Zhipu) | 2026-09-27 | 2026-09-23 | $2.80 / $8.80input / output per 1M tokens | 1 |
| Qwen3.8 Max Prime | Qwen (Alibaba) | 2026-09-27 | 2026-09-23 | $4.00 / $12.00input / output per 1M tokens | 1 |
| Command A+ | Cohere | 2026-09-22 | 2026-09-22 | $0.30 / $1.50input / output per 1M tokens | 1 |
| Qwen3.8 Omni Flash | Qwen (Alibaba) | 2026-09-22 | 2026-09-21 | $0.15 / $0.47input / output per 1M tokens | 0 |
| GPT-6 Luna | OpenAI | 2026-09-22 | 2026-09-22 | $0.10 / $0.50input / output per 1M tokens | 3 |
| GPT-6 Luna Pro | OpenAI | 2026-09-22 | 2026-09-22 | $0.10 / $0.50input / output per 1M tokens | 2 |
| GPT-6 Sol | OpenAI | 2026-09-22 | 2026-09-22 | $2.00 / $10.00input / output per 1M tokens | 3 |
| GPT-6 Sol Pro | OpenAI | 2026-09-22 | 2026-09-22 | $2.00 / $10.00input / output per 1M tokens | 2 |
| Claude Opus 5.5 | Anthropic | 2026-09-22 | 2026-09-22 | $4.00 / $20.00input / output per 1M tokens | 5 |
| MiMo-V2.6-Flash | Xiaomi | 2026-09-21 | 2026-09-21 | $0.10 / $0.28input / output per 1M tokens | 7 |
| Grok 4.7 | xAI | 2026-09-21 | 2026-09-21 | $2.00 / $6.00input / output per 1M tokens | 1 |
Week of 2026-09-14
| Model | Developer | First seen (UTC) | Released | Headline price | Providers |
|---|---|---|---|---|---|
| Lyria 3.5 | Google DeepMind | 2026-09-20 | 2026-09-19 | $0.10 per generation | 1 |
| GLM 5.3 FlashX | Z.ai (Zhipu) | 2026-09-19 | 2026-09-18 | $0.37 / $1.25input / output per 1M tokens | 1 |
What is listed
The same significance rules as the SF·SIX feed: a release needs an identified developer and corroboration (a known lab’s model listed on the OpenRouter market, on the lab’s own channel or with two providers; any other developer’s model with five independent providers). Adapters (LoRAs), merges, quantizations, fine-tunes, moving aliases, community uploads and late imports of older models are left out; they stay listed in the model directory. “First seen” is when SF·SIX first observed the model at a source (not the developer’s announcement); “Released” is the date the source reports. Models already listed when SF·SIX began recording first sightings (2026-09-16) are not counted. Endpoints of one model and its batch or free variants are one row. Prices are the headline standard price (cheapest qualifying provider) or the cheapest native unit price.
Frequently asked questions
Which AI models were released this week?
DeepSeek V4.1 Flash 1017, Bytedance Omnihuman V2, FLUX.3 Klein Image Editing, NVIDIA Nemotron 3.5 Super Text, Muse Spark Open, Nemotron 3.5 Super VL, Gemini 4 Argon, Muse Wonder, first listed on SF·SIX since 2026-10-12 (UTC).
What counts as a new model here?
A model first seen by SF·SIX in the period whose developer is identified and that is corroborated: a known lab’s model listed on the OpenRouter market, published on the lab’s own channel or served by two providers, or any other developer’s model served by at least five independent providers. Adapters (LoRAs), merges, quantizations, fine-tunes, moving aliases and models released more than 30 days before they appeared are not counted.